Journal

Hong Kong J Radiol 2025 Dec;28(4):e232-3

 

 

Department of Radiology, Hong Kong Children’s Hospital, Hong Kong SAR, China

 

  Full Article in PDF

The introduction of cross-sectional imaging during procedures has greatly improved treatment precision and allowed real-time assessment of therapeutic outcomes, particularly in the realm of interventional oncology. The first hybrid angiography–computed tomography (angio-CT) system was developed in the early 1990s by Professor Yasuaki Arai at Aichi Medical Centre in Japan.[1] In its early stages, integration between the two modalities was minimal, with each operating largely independently and requiring the operator to manually combine the imaging data.[2] The development of flat-panel detectors in the late 1990s paved the way for the rapid adoption of cone beam CT (CBCT). Compared to angio-CT, CBCT has since become the more widely utilised modality,[3] largely attributable to the higher cost and infrastructural demands of angio-CT systems, particularly the need to accommodate a sliding gantry. However, recent advances in workflow efficiency, decreasing relative costs, and the multipurpose capabilities of angio-CT systems have led to renewed interest. Increasingly, institutions are accepting the higher upfront investment, recognising the potential long-term benefits in terms of improved patient outcomes and enhanced procedural room utilisation.

 

In this issue of the Hong Kong Journal of Radiology, Wong et al[4] presented a case-based review that explores the advantages of angio-CT technology through a series of illustrative examples. While most of the interventional radiology (IR) literature has focused primarily on oncologic applications, this review highlights its broader utility in various vascular interventions, including embolotherapy for acute haemorrhage, management of pulmonary arteriovenous fistulae, and adrenal venous sampling. Additional potential non-oncological applications—though not addressed in this review— include complex drain placements, acute trauma management, prostate artery embolisation, geniculate artery embolisation, complex venous reconstruction, and lymphatic interventions.[5] [6] These examples underscore the versatility of angio-CT systems across a wide spectrum of IR procedures.

 

The most significant advantage of angio-CT lies in its ability to integrate high-resolution CT imaging with selective angiography and fluoroscopy, thereby eliminating the need to transfer patients to a separate CT scanner during critical procedural steps. Motion and beam-hardening artefacts commonly encountered with CBCT are minimised with angio-CT due to its superior temporal resolution. The capacity to accurately delineate perfused tissue volume is particularly valuable in procedures where extra-target embolisation may have serious consequences, such as geniculate artery embolisation and prostate artery embolisation. Additionally, the ability to provide three-dimensional needle guidance is invaluable for complex interventions such as sharp recanalisation in venous reconstructions and antegrade percutaneous puncture of the cisterna chyli. In the setting of acute polytrauma and stroke, angio-CT enables both diagnostic CT and subsequent therapeutic embolisation to be performed within the same gantry, thereby saving precious time and reducing the risks associated with transferring critical, often haemodynamically unstable, patients from one venue to another.

 

Converting a conventional IR suite into a hybrid angio-CT suite has been shown to enhance operational efficiency across both IR and diagnostic radiology services. This conversion allows CT scanners, which previously had to be shared between diagnostic and interventional services, to be dedicated solely to diagnostic imaging, while enabling CT-guided procedures to be performed directly within the hybrid IR suite. For instance, one institution reported a 19.1% relative increase in the utilisation of formerly shared CT scanners for diagnostic imaging, along with a 287.1% relative increase in the use of the hybrid IR suite, compared with the overall growth rates of both diagnostic radiology and IR departments.[7] [8] In addition, the potential use of the angio-CT scanner as a diagnostic scanner affords scheduling flexibility for both patients and physicians and contributes to workflow efficiency and optimisation.

 

A major limitation of angio-CT systems is the substantially higher cost of installation compared with single-modality fluoroscopy suites. In addition, these systems require a larger physical footprint, typically at least 50 m², to accommodate both CT and fluoroscopy units. To date, robust quantitative evidence demonstrating improved patient outcomes or cost-effectiveness is lacking. Comparison of radiation dose between CBCT and angio-CT is also challenging due to variability in imaging protocols and technical parameters, which introduces uncertainties in direct comparisons. Notably, a CT acquisition during a procedure may reduce the need for multiple digital subtraction angiography runs, thereby lowering and more evenly distributing overall radiation exposure to the patient. While one study found no significant difference in effective dose per CT scan between CBCT and angio-CT,[9] limited available data suggest that angio-CT systems may reduce the overall effective dose from both CT and angiographic imaging compared with CBCT systems.[10]

 

Hybrid angio-CT systems offer institutions a valuable opportunity to expand treatment capabilities and optimise workflow. The integration of diagnostic-quality intraprocedural CT with conventional angiography and fluoroscopy has substantially broadened the scope of interventional procedures while improving room utilisation. While the synergistic benefits of angio-CT are clear, the decision to invest in such a system should be informed by a comprehensive needs assessment and institutional considerations, including case mix, procedural complexity, patient volume, and the potential to enhance workflow efficiency and translate into improved patient outcomes.

 

 

 

 

 

 

 

 

 

 

 

 

Hong Kong J Radiol 2025 Dec;28(4):e234-42 | Epub 17 November 2025

 

 

¹ Department of Diagnostic and Interventional Radiology, Kwong Wah Hospital, Hong Kong SAR, China

² Department of Radiology, Hong Kong Children’s Hospital, Hong Kong SAR, China

 

  Full Article in PDF

Renal angiomyolipoma (AML) is the most common benign solid renal tumour.[1] The majority (approximately 80%) occur sporadically, while the rest (approximately 20%) are associated with phakomatoses, most commonly tuberous sclerosis.[2] AML belongs to the family of tumours with perivascular epithelioid cellular differentiation.[3] It typically contains both angiomyogenic and fatty components, with the latter readily identifiable in computed tomography (CT) due to its hypoattenuating nature (< -10 Hounsfield unit [HU]) [Figure 1].[4] [5]

 

Figure 1. Renal angiomyolipoma (AML) pre- and post–prophylactic embolisation. (a) Preprocedural contrast-enhanced computed tomography (CT) of the abdomen with coronal reformation shows an exophytic mass arising from the upper pole of the right kidney containing both soft tissue (cross) and fat (triangle) components, suggestive of an AML (arrow). (b) Pre-embolisation catheter angiogram demonstrates hypervascular nature of the AML (arrow) with tortuous vessels. (c) Selective catheter angiogram via microcatheter identifies the feeding artery of the AML (arrow), allowing targeted deployment of embolisation agent. (d) Post-embolisation catheter angiogram confirms technical success with complete resolution of tumour stain. (e) Postprocedural contrast-enhanced CT of the abdomen shows reduction in total volume of AML (arrow), with greater shrinkage of the angiomyogenic (cross) than fatty (triangle) components.

 

It is well recognised that AML carries a risk of rupture with bleeding, especially for larger tumours, which can lead to fatal consequences.[6] [7] Treatment options include transcatheter arterial embolisation and radiofrequency ablation, as well as partial or radical nephrectomy.[8] Selective arterial embolisation can be performed in an emergency setting for AML with active bleeding. It can also be a prophylactic treatment to reduce tumour size and its risk of haemorrhage (Figure 1).[9] [10] It has been suggested that for AML of 4 cm or above in diameter, or those with microaneurysms 0.5 cm or above in the feeding artery, prophylactic selective arterial embolisation is indicated.[11] Different embolisation agents have been reported in the literature, including microparticles, such as microspheres, and liquid agents, such as ethanol. Past studies have shown that prophylactic embolisation could reduce the size of AML, thus reducing its haemorrhagic risk.[9] [10] [12] In addition, the risk of haemorrhage is mainly attributed to the angiomyogenic component of AML.[13] [14] [15] Yet, there are limited studies accurately assessing how tumour composition changes after treatment.

 

For AML, tumour size has been shown to be associated with risk of spontaneous rupture, with larger ones more likely to bleed.[6] [7] [11] This study therefore aimed to assess the efficacy of prophylactic selective arterial embolisation in determining the rupture risk post-embolisation and reducing the volume of AML using semi-automatic segmentation as a measurement tool. Changes in its angiomyogenic and fatty components were also evaluated.

 

This was a single-centre, single-arm retrospective study. Consecutive adult patients (≥18 years old) who underwent prophylactic embolisation of AML (Figure 2) in a public acute general hospital between January 2009 and January 2024 were included. Exclusion criteria included paediatric patients (<18 years old), patients who underwent emergency embolisation of ruptured AML, and patients without pre- or postprocedural CT.

 

Figure 2. Protocol for prophylactic embolisation of renal angiomyolipoma.

 

Clinical data of the included patients were retrieved from the radiology information system of the hospital network. They included demographics and medical history, such as tuberous sclerosis status. Presenting symptoms and postprocedural complaints were recorded. Pre- and post-intervention blood tests, such as haemoglobin level and renal function tests, were documented.

 

Details of prophylactic embolisation of AML were logged. They encompassed the type and amount of embolisation agents deployed, as well as catheters and guidewires used, which were chosen based on the operators’ preference. Data on technical success, defined as complete angiographic resolution of tumour stain and microaneurysms, as well as contrast stasis of the feeding artery, were documented. Intraoperative and immediate postprocedural complications were recorded. Subsequent clinical follow-up was reviewed for post-embolisation tumour rupture.

 

Pre- and post-embolisation plain and contrast-enhanced CTs of the abdomen in DICOM (Digital Imaging and Communications in Medicine) format were obtained from the picture archiving and communication system of the hospital network. The time interval between the day of CT examination and interventional procedure was logged. DICOM images were assessed using 3D Slicer (macOS version 5.6.2; The Slicer Community), an open-source image computing platform.[16] Semi-automatic segmentation of AMLs was performed in the following sequence (Figure 3): (1) reformation of contrast-enhanced CT images in axial, coronal and sagittal planes; (2) manual contouring of tumour and non-tumour regions on limited CT slices (<5); (3) automatic segmentation of tumour and non-tumour regions using the ‘grow from seeds’ algorithm; (4) manual refinement of segmented regions using ‘paint’ and ‘erase’ algorithms; (5) automatic differentiation between angiomyogenic and fatty components of AML using a ‘threshold’ algorithm, with the threshold set at ≥ -10 HU for the angiomyogenic component and < -10 HU for the fatty component; (6) automatic volume rendering of tumour and non-tumour regions; and (7) automatic volumetric computation of the entire AML, as well as its angiomyogenic and fatty components. In addition, laterality and polarity of AML, as well as the presence or absence of aneurysms, were documented. In postprocedural CT, any complications, including renal parenchymal infarction, haematoma, abscess, pyelonephritis, or hydronephrosis, were recorded.

 

Figure 3. Semi-automatic segmentation of renal angiomyolipoma (AML) using 3D Slicer. (a) Digital Imaging and Communications in Medicine images of contrastenhanced computed tomography (CT) of the abdomen are reformatted in the axial (upper left), coronal (lower left), and sagittal planes (lower right). An AML (arrows) arises from the upper pole of the left kidney. (b) Contrast-enhanced CT of the abdomen reformatted in sagittal planes. Contours of AML (arrows) and other non-tumour regions are drawn manually on several (<5) CT slices. (c) Automatic segmentation of AML (arrows) is performed in the axial (upper left), coronal (lower left), and sagittal planes (lower right) using the ‘grow from seeds’ algorithm. Further refinement of the segmented regions is possible using the ‘paint’ and ‘erase’ algorithms. Automatic three-dimensional rendering of (d) non-tumour regions and (e) the AML are shown. Further volumetric analysis, such as determining the angiomyogenic and fatty components of the AML using -10 Hounsfield unit as the threshold, can be performed.

 

Statistical analysis was performed using SPSS (macOS version 29.0; IBM Corp, Armonk [NY], United States). The distribution of all numerical data was first tested for normality using the Shapiro-Wilk test. The Wilcoxon signed-rank test was used to compare paired data, such as pre- and postprocedural volumetric changes in each AML. The Mann-Whitney U test was used for comparison between unpaired data. Spearman’s rank correlation coefficient was used to identify association between variables. A p value of < 0.05 was considered statistically significant.

 

This manuscript was prepared in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.

 

There were 31 prophylactic embolisations of AMLs in adult patients performed from January 2009 to January 2024. Five patients underwent emergent embolisation of ruptured AMLs. One patient did not have preoperative CT available for assessment. These six patients were therefore excluded. A total of 25 prophylactic embolisations of AML in 22 patients were finally included in the study, of which three patients had repeated embolisation (n = 3, 13.6%) [Figure 4]. The median age of the patients on the day of embolisation was 60.0 years (interquartile range, 15.0). Most patients were female (n = 18, 81.8%). There was no patient with tuberous sclerosis. Four patients (18.2%) had known chronic kidney disease due to diabetic nephropathy (n = 2, 9.1%), hypertensive nephropathy (n = 1, 4.5%) and IgA nephropathy (n = 1, 4.5%) [Table 1].

 

Figure 4. Selection of study population with prophylactic embolisation of renal angiomyolipoma.

 

Table 1. Demographics and clinical information of the study population.

 

All prophylactic embolisations of AMLs were performed due to large tumour size (≥4 cm in diameter). In one case (4.0%), there was a 0.5-cm aneurysm identified in preprocedural assessment, which was successfully embolised. Microspheres (Embosphere Microsphere; Merit Medical Systems, South Jordan [UT], United States) were employed in two-thirds of all interventions (n = 17, 68.0%). The sizes of the microparticles ranged between 100-300 μm (n = 2, 11.8%), 300-500 μm (n = 6, 35.3%) and 500-700 μm (n = 9, 52.9%). Ethanol was used in the remaining one-third of the cases (n = 8, 32.0%). Ethanol was radio-opacified with ethiodised oil in a ratio of 7:3 for embolisation of the other cases. Selective arterial embolisation with microcatheters and microguidewires was performed in every case. All prophylactic embolisation achieved technical success. There were no intra-procedural or immediate postprocedural complications encountered. The median time intervals between pre- and postprocedural CT with prophylactic embolisation were 90.0 days and 107.0 days, respectively. Postprocedural CT showed a small (2.0 cm in diameter) subsegmental renal infarction in one case (4.0%). No other complications were seen. There were no significant changes in haemoglobin level or renal function tests before and after prophylactic embolisation. Median clinical follow-up duration was over 4 years (49.0 months), with a minimum of 6 months. There were no post-embolisation rupture of AML (Table 2).

 

Table 2. Details of prophylactic embolisation of renal angiomyolipoma (n = 25).

 

The median total volume of AMLs in pre- and postprocedural CTs were 67.5 cm³ and 35.7 cm³, respectively, showing significant interval shrinkage, with 41.4% total tumour volume reduction. Both angiomyogenic and fatty components showed significant interval reduction in size, attaining 73.6% and 14.0% volume loss, respectively. The angiomyogenic component of the AMLs showed significantly greater reduction in size compared to the fatty component (Table 3).

 

Table 3. Compositions of renal angiomyolipoma before and after prophylactic embolisation (n = 25).

 

Correlation analysis revealed AMLs with a greater proportion of angiomyogenic component and smaller proportion of fatty component in preprocedural CT were associated with greater tumour volume reduction after prophylactic embolisation. No other clinical or procedural factors associated with total tumour shrinkage were identified (Table 4).

 

Table 4. Correlation analysis performed to identify factors associated with total angiomyolipoma volume reduction after prophylactic embolisation (n = 25).

 

Prophylactic embolisation of AML with either microspheres or ethanol achieved significant reduction in total tumour size, with 26.2% (p < 0.001) and 42.7% (p = 0.008) volume loss, respectively. Using either embolisation agent, there were significant reduction in volume of both angiomyogenic (microspheres: 71.6%, p < 0.001; ethanol: 81.0%, p = 0.008) and fatty components (microspheres: 12.7%, p < 0.001; ethanol: 29.7%, p = 0.008), with the angiomyogenic component showing significantly greater volume loss than the fatty component using either embolisation agent (both p < 0.001). Comparing microspheres and ethanol, there were no statistically significant differences in their efficacy of reduction of the total tumour volume, angiomyogenic or fatty components of AML (Table 5).

 

Table 5. Efficacy of employing microspheres versus ethanol in prophylactic embolisation of renal angiomyolipoma.

 

Transcatheter embolisation of AML is recognised as a safe intervention.[9] [10] Compared to more invasive treatment options such as partial or radical nephrectomy, transcatheter selective arterial embolisation typically only requires local anaesthesia, has lower risks of bleeding and infection, and allows shorter admission times. Some authors therefore suggest transcatheter embolisation as the first-line treatment option.[8] In our study, no intra-procedural or immediate complications were encountered. However, in one patient, a small subsegmental renal infarction was seen in postprocedural CT. This highlights the importance of follow-up imaging, which encompasses assessment of treatment efficacy, as well as identification of complications.

 

There was significant change in tumour volume after prophylactic embolisation of AML, achieving over 40% reduction in median total volume amongst our study population. With decreased tumour volume, the risk of spontaneous haemorrhage would be lowered.[6] [11] It was reassuring that none of the included patients encountered post-embolisation tumour rupture in clinical follow-up with a median duration of over 4 years. These findings demonstrate that prophylactic embolisation of AML is a safe and effective means to reduce haemorrhagic risk, concurrent with previous studies.[9] [10]

 

Various materials for prophylactic embolisation of the kidney have been suggested in the literature, with microspheres and ethanol being two of the most commonly adopted agents. In this study, both microspheres and ethanol effectively reduced the size of AMLs by over 25%, without a statistically significant difference between the two agents. To the best of our knowledge, there is no large-scale study establishing whether microspheres or ethanol is the superior prophylactic embolisation agent for AML.[9] [10] [12] In our centre, this choice depended on the operators’ preference.

 

It has been proposed that the effectiveness of prophylactic embolisation in achieving volume reduction depends on the composition of the AML, which has variable angiomyogenic and fatty components. The angiomyogenic component usually demonstrates greater response to embolisation due to its vascular nature, whereas the fatty component is hypovascular and more treatment-resistant.[9] [17] A study by Han et al[17] showed near-complete resolution of the angiomyogenic component after prophylactic embolisation, but the fatty component only partially shrank. In their study, the proportion of angiomyogenic and fatty components were evaluated on a transverse image at the middle of the tumour. However, this might not reflect the actual composition of the entire AML. In our study, semi-automatic segmentation was performed, and the angiomyogenic and fatty components were differentiated using -10 HU as the threshold. This allowed a more accurate volumetric assessment of AML. Similar to prior studies, there was significantly greater reduction in the angiomyogenic component than the fatty component after embolisation.

 

AMLs with a greater proportion of angiomyogenic component and smaller proportion of fatty component are associated with greater total volume reduction after embolisation. For AML with high fatty content, patients and clinicians may be concerned that about the smaller postprocedural volume reduction. However, the angiomyogenic component of AML, which is the main culprit in haemorrhage, has shown good response to embolisation.[9] [17] In our study, the angiomyogenic component achieved over 70% volume reduction, which could be reassuring to both patients and clinicians.

 

First, none of the patients in our study had tuberous sclerosis. Treatment efficacy for sporadic and tuberous sclerosis–associated AML may differ and have not been explored. Second, there was a lack of a control group to compare rupture risk in patients who received prophylactic embolisation versus those who did not. A double-arm study could better assess treatment effect. Third, the sample size was limited. This may be partly attributed to the relatively low prevalence of AML, which is below 0.5% in the population.[18] A multi-centre study with larger sample sizes is a potential future direction.

 

Prophylactic embolisation of AML effectively reduced tumour volume, with more significant changes in the angiomyogenic component compared to the fatty component. No rupture or haemorrhage was documented post-embolisation with a median follow-up of over 4 years.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hong Kong J Radiol 2025 Dec;28(4):e268-75 | Epub 9 December 2025

 

 

Department of Diagnostic and Interventional Radiology, Kowloon West Cluster, Hong Kong SAR, China

 

  Full Article in PDF

Infective endocarditis (IE) affects 1.7 to 6.2 individuals per 100,000 population per year and remains a life-threatening condition.[1] Staphylococci are the most frequent causative organisms.[1] [2] Neurological complications are the most common and severe extracardiac complications of IE[3] and have been reported as the presenting symptom in up to 47% of cases.[4] These complications are caused by cerebral septic embolisation of endocardial vegetations. Patients with neurological complications have significantly higher mortality compared to those without (24% vs. 10%; p < 0.03).[3] Neuroimaging leads to the identification of valvular surgery indications in about 22% of patients with symptoms of neurological complications of IE, and in 19% of asymptomatic IE patients.[2] Up to 82% of patients have cerebral lesions on magnetic resonance imaging (MRI) performed within 7 days after admission.[1] MRI findings influence diagnostic classification and other clinical decisions in 28% of patients, including modification of medical or surgical treatment plans.[5] According to the 2015 European Society of Cardiology guidelines,[6] the presence of cerebral emboli in patients with left-sided valvular vegetations greater than 10 mm is an indication for urgent valve surgery to prevent further embolisms. Familiarity with the neurological imaging findings is essential for early diagnosis of this complication of IE, allowing a window for early and specific treatment, thereby reducing mortality. However, the wide spectrum of presentations on neuroimaging poses diagnostic challenges to radiologists, especially when cerebral septic embolism is the first presentation. This pictorial essay aims to review the spectrum of presentations and the use of multimodality imaging to increase awareness of the classic diagnostic imaging findings of cerebral septic emboli secondary to IE, and to highlight the role of interventional radiology in clinical management.

 

The diagnosis of IE is made according to the Modified Duke criteria,[1] which include the presence of major arterial emboli, mycotic aneurysm, and intracranial haemorrhage as part of its minor criteria.

 

Computed tomography (CT) is the first-line imaging study in patients with neurological symptoms as it is readily available. MRI, including susceptibility-weighted imaging (SWI) and diffusion-weighted imaging (DWI), is required to detect more subtle findings such as cerebral microbleeds and early infarcts. Further investigations with computed tomographic angiography (CTA) or magnetic resonance angiography (MRA) are useful for detecting mycotic aneurysms, while digital subtraction angiography (DSA) remains the gold standard and should be performed in clinically suspicious cases with negative CTA or MRA.[7] There is growing support for performing screening MRI in patients with suspected or confirmed IE, given the frequency of asymptomatic findings and its usefulness in decision-making. However, its cost-effectiveness and impact on mortality reduction remain to be seen.[8]

 

Neurological complications of IE may present as cerebral infarcts, micro- or macro-haemorrhage, abscess, and meningitis. The pooled frequency of individual findings on MRI is as follows: acute ischaemic lesions (61.9%), cerebral microbleeds (52.9%), macro-haemorrhages (24.7%), abscess or meningitis (9.5%), and intracranial mycotic aneurysm (6.2%).[8] Accurate identification of these lesions allows early diagnosis of IE complications and individualised management strategies.

 

Ischaemic stroke is the most common neurological manifestation of IE. It can result from embolisation of endocardial vegetations, leading to occlusion of intracerebral arteries.[3] The incidence of cerebral ischaemia is correlated with left-sided endocarditis (especially involving the anterior mitral valve leaflet), larger endocardial vegetation size (>10 mm), mobile vegetations, and Staphylococcus aureus infection.[3]

 

Disseminated ischaemic lesions may result from multiple emboli occurring over a short period or fragmentation of an embolus in the heart or aorta. The presence of multiple cortical and subcortical cerebral infarcts of varying ages within different vascular territories (especially watershed areas) or bihemispheric involvement suggests the diagnosis of septic emboli[4] (Figure 1). Large emboli tend to cause cortical infarction in the middle cerebral artery (MCA) territory, while smaller emboli often lodge distally in terminal cortical branches of the anterior cerebral artery and MCA, resulting in small peripheral infarcts at the grey-white matter junction.[9] It is worth noting that isolated brainstem strokes are rarely caused by cardioembolism.

 

Figure 1. A 69-year-old woman presented with fever and confusion 3 days after dental surgery. (a) Axial T2-weighted magnetic resonance imaging (MRI) shows multiple foci of hyperintense signals involving the bilateral high frontal and parietal cortices (asterisks). (b) Axial diffusion-weighted imaging (DWI) with a high b-value revealed corresponding hyperintense signal (arrowheads) with low signal on apparent diffusion coefficient map (ADC) [not shown], suggestive of restricted diffusion. (c) Axial T2-weighted MRI shows hyperintense signals involving the bilateral basal ganglia, capsular regions, and thalami (asterisks). (d) Axial DWI with a high b-value shows patchy areas of hyperintense signal (arrowheads), with low signal on the ADC map (not shown), suggestive of restricted diffusion. These findings were suggestive of acute infarcts. The distribution pattern raised suspicion for an embolic shower, involving both deep perforating arteries and cortical branches. Consequently, an echocardiogram was performed and revealed a ventricular septal defect and tricuspid valve vegetation, consistent with a paradoxical embolism. The patient was managed conservatively with intravenous antibiotics and showed good neurological recovery.

 

DWI is useful for assessing the temporal relationship of ischaemic lesions. Acute infarcts appear as hyperintense on DWI and hypointense on apparent diffusion coefficient mapping. Over time, the apparent diffusion coefficient signal increases and pseudonormalises in about 1 week, signal increases and pseudonormalises in about 1 week, while the DWI signal decreases and pseudonormalises in about 2 weeks.[10]

 

Cerebral abscesses and meningitis are uncommon neurological manifestations of IE, occurring in up to 9.5% of patients.[8]

 

Typically, multiple abscesses appear in the MCA territory at the grey-white matter junction, often with vasogenic oedema and associated mass effect or haemorrhage.[3] On CT, cerebral abscesses are usually hypodense with ring enhancement, but MRI is more sensitive. Classic MRI features include lesions that are hypointense on T1-weighted images and hyperintense on T2-weighted images, with ring enhancement and central restricted diffusion (Figure 2). A dual rim sign, two concentric rims surrounding the abscess cavity, where the outer rim is hypointense and the inner is relatively hyperintense, may be visible on SWI or T2-weighted imaging. Cerebral abscesses may also arise near mycotic aneurysms (Figures 3 and 4). The presence of leptomeningeal enhancement on MRI or CT can suggest concomitant meningitis.

 

Figure 2. A 56-year-old woman presented with fever and a bilateral lower extremity rash. Physical examination revealed a pansystolic murmur with radiation to the left axilla and splinter haemorrhages. Echocardiography demonstrated mitral valve regurgitation and prolapse with vegetation. Blood culture yielded Streptococcus sanguinis. (a) Axial post-contrast T1-weighted image shows a small ring-enhancing lesion in the right basal ganglia (arrowhead). The lesion is hyperintense on T2-weighted image (not shown). There is associated focal leptomeningeal enhancement in the adjacent right frontal lobe cortex (arrow), suggestive of leptomeningitis. (b) Axial diffusion-weighted imaging with a high b-value demonstrates a focal central hyperintense signal (arrow) within the previous right basal ganglia ring-enhancing lesion. (c) The corresponding apparent diffusion coefficient map shows hypointense signal (arrow). These findings were indicative of restricted diffusion and therefore consistent with an abscess. (d) Axial gradient echo sequence reveals a focus of susceptibility artefact in the left occipital lobe (arrowhead), consistent with a microbleed. All findings resolved with conservative management with a 6-week course of broad-spectrum antibiotics. Mitral valve replacement was proposed but declined by the patient.

 

Figure 3. A 27-year-old woman presented with fever, left-sided weakness, and slurred speech. (a) Initial axial non-contrast computed tomography (CT) of the brain shows a small focus of hypodensity in the right temporoparietal region (arrow). (b) Follow-up axial non-contrast CT 1 day later demonstrates rapid enlargement of the hypodensity in the right temporoparietal region (arrow), with a small acute haemorrhagic focus (arrowhead). (c) Axial computed tomography angiography (CTA) shows a small contrast-enhancing focus at the proximal M2 segment of the right middle cerebral artery (arrowhead) within the infarct, suggestive of an aneurysm. (d) Volumetric rendering of the CTA shows the saccular morphology of the proximal right M2 aneurysm (arrowhead), consistent with a mycotic aneurysm.

 

Figure 4. Same patient as in Figure 3. (a) Axial post-contrast T1-weighted magnetic resonance image shows a ring-enhancing lesion in the right temporoparietal region (arrow), with a central non-enhancing area and a peripheral enhancing focus (arrowhead). (b) Axial diffusion-weighted image with a high b-value demonstrates central hyperintense signal (arrow). (c) The apparent diffusion coefficient map shows corresponding hypointense signal (arrow). Overall findings are suggestive of a cerebral abscess with a mycotic aneurysm. Subsequent echocardiography revealed mitral valve regurgitation and prolapse with vegetation. Blood culture was negative. The patient underwent burr hole drainage of the abscess, and pus culture yielded Staphylococcus aureus. She was subsequently treated with a course of intravenous vancomycin.

 

Macrohaemorrhage usually results from haemorrhagic transformation of ischaemic stroke, progression of microhaemorrhages, or rupture of mycotic aneurysms. Haemorrhagic transformation occurs more frequently in embolic strokes (51%-71%) than in non-embolic strokes (2%-21%)[11] and may present as petechial haemorrhage or large parenchymal haematomas.[9] Cerebral ischaemic lesions of varying ages across multiple vascular territories and different haemorrhagic patterns would raise suspicion for cardiac emboli. In the context of underlying IE, cerebral septic embolism is a likely diagnosis (Figure 5).

 

Septic emboli damage the endothelium and disrupt the blood-brain barrier, resulting in inflammatory vasculitis and small vessel rupture, often leading to cerebral microbleeds or even intracerebral haemorrhage. One study found that cerebral microbleeds in 57% of patients with IE.[4] These microbleeds appear as hypointense foci on T2^* imaging or SWI MRI often in the cortex, and less frequently in subcortical white matter, basal ganglia, or posterior fossa.[4]

 

Figure 5. A 62-year-old man with known mitral valve regurgitation. (a) Axial diffusion-weighted imaging of the brain with a high b-value shows a focal hyperintense signal in the left frontal lobe (arrowhead), with corresponding hypointensity on the apparent diffusion coefficient map (not shown). (b) Associated susceptibility artefact is noted in the same region (arrowhead), suggestive of haemorrhagic transformation. (c) Axial unenhanced computed tomography (CT) of the brain 1 month later shows a new haemorrhagic infarction in the right occipital lobe with mild perilesional oedema (arrow). (d) Axial computed tomography angiography (CTA) of the brain shows a tiny contrast-enhancing focus within the haemorrhagic infarct, suggestive of an aneurysm (arrowhead), while the left frontal infarct shows signs of chronicity (arrow). (e) Axial non-contrast CT of the brain 1 day later shows a new right frontal lobe infarct, as well as new subarachnoid haemorrhage in the right frontal region (arrowhead) and suprasellar cistern (arrow). (f) Axial non-contrast CT of the brain 3 days later reveals a new right parieto-occipital lobe infarct (arrow). (g) Axial non-contrast CT of the brain 1 month later shows a new haemorrhagic infarct in the high left parietal lobe (arrow). (h) Axial CTA of the brain shows a tiny contrast-enhancing focus (arrowhead) within the haemorrhagic infarct suggestive of aneurysm. The presence of multiple haemorrhagic infarcts of different timing, some with associated aneurysms, is highly suggestive of cerebral septic emboli with mycotic aneurysms. Subsequent echocardiography revealed mild mitral and tricuspid regurgitation with vegetations on both mitral valve leaflets. Blood cultures yielded Rothia dentocariosa. A diagnosis of infective endocarditis was established and the patient was treated with intravenous antibiotics.

 

Cerebral septic emboli can trigger inflammation and weakening of vessel walls, forming mycotic aneurysms.[7] These aneurysms are found in about 6.2% of patients with IE and may shrink, enlarge, or develop de novo within 1 week to 3 months of starting antibiotics.[12] Mycotic aneurysms have a 2% to 10% risk of rupture regardless of their size and are associated with a high mortality rate of 80%.[7] About 22% of IE patients presenting with intracerebral haemorrhage have mycotic aneurysms which should be promptly identified.[13] CTA or MRA should be performed to confirm the diagnosis (Figures 5 and 6), followed by DSA for clear delineation of the number, size and location of the mycotic aneurysms and surgical or endovascular planning.

 

Figure 6. Same patient as in Figure 5. (a) Axial plain computed tomography of the brain 2 weeks after antibiotic therapy shows a new, large haemorrhagic infarct in the right frontal lobe with mass effect (arrow). (b) Axial unenhanced computed tomography at a lower level from the same study demonstrates a preexisting haematoma with intraventricular extension (arrow). (c) Volumetric rendering of computed tomography angiography reveals a saccular aneurysm arising from a distal branch of the right anterior cerebral artery (arrowhead), saccular aneurysms of the right middle cerebral artery (asterisk), and a left posterior cerebral artery aneurysm (arrow). The peripheral location and saccular morphology of these aneurysms are highly suggestive of mycotic origin.

 

CTA and MRA have low sensitivity for small (<5 mm) or distal mycotic aneurysms.[1] Aneurysms near the skull base may be overlooked on CTA, while those in low-flow areas may be missed on time-of-flight MRA.[14] In cases with clinical suspicion of mycotic aneurysm but negative CTA or MRA, DSA should be performed.[1]

 

Features favouring mycotic aneurysms include multiplicity, saccular shape, distal location (such as MCA segments 2 to 4 or posterior cerebral artery), size or morphological changes on consecutive angiograms, presence of other intra- or extra-cranial mycotic aneurysms, adjacent arterial occlusion or stenosis, and cerebral infarction at the aneurysm site[13] (Figure 6).

 

Neurological complications from IE are life-threatening and require multidisciplinary management, involving neurosurgeons, radiologists, cardiologists, and microbiologists. Empirical intravenous antibiotic therapy is promptly administered and later tailored according to culture sensitivity results. Valvular replacement combined with antibiotics yield better outcomes than antibiotics alone in left-sided endocarditis.[15]

 

Radiologists play a key role in both diagnosis and guiding treatment by accurately reporting the type and severity of each lesion. Surgical drainage can be considered in cases of cerebral abscesses with significant mass effect. Antiplatelet drugs and anticoagulants are contraindicated in both ischaemic stroke and macrohaemorrhage caused by septic embolism due to the high risk of bleeding.[3] Cardiac surgery should be postponed for at least 4 weeks after a clinically significant intracranial haemorrhage or large ischaemic infarct.[3] Mycotic aneurysms should be excluded before open heart surgery for valvular replacement requiring anticoagulation to reduce bleeding risk.[15]

 

Interventional radiologists play an evolving role of in treating mycotic aneurysms in collaboration with neurosurgeons. Techniques include preoperative CTA or MRA with volumetric rendering, road-map technique for neuro-navigation, and cone beam CTA for postprocedural monitoring. Given the unpredictable nature of mycotic aneurysms and the weak correlation between size and rupture risk, surgical or endovascular treatment should be considered for unruptured aneurysms that enlarge or do not regress on follow-up imaging.[7] [14] Ruptured or symptomatic mycotic aneurysms also require surgical or endovascular intervention.[14] A surgical approach is indicated when an aneurysm exerts mass effect[14] or supplies an eloquent brain region.[1] However, clipping may be difficult due to a wide or absent aneurysmal neck and fragile vessels.[3]

 

An endovascular approach is indicated for those unfit for surgery due to cardiac disease.[3] It can be divided into direct or indirect approaches. An indirect approach with parent artery occlusion is the endovascular treatment of choice, especially for distally located aneurysms and circumferential vessel involvement. However, parent artery sacrifice is not possible at times and the direct approach may remain the only viable option. The direct approach using coils or liquid embolic agents allows precise control of the aneurysm while preserving distal flow from the parent artery. Endovascular coiling may be a safer option with higher occlusion and lower procedure-related complication rates[7] (Figure 7). Detachable coils allow precise deployment and better durability compared with liquid embolic agents. They are preferred in proximal aneurysms, while liquid embolic agents are more suitable for distal aneurysms not accessible by microcatheter. Intracranial flow diverters can be used to divert turbulent blood flow from the aneurysm and preserve laminar blood flow in the main vessel and its side branches. With reduced blood flow to the aneurysm and gradual vessel remodelling, this results in progressive aneurysmal sac thrombosis[16] (Figure 8). It is important to note that mycotic aneurysms may grow after simple coiling, while the parent artery may thrombose after flow diverter placement in the setting of infection.

 

Figure 7. The same patient as Figures 5 and 6 underwent surgical clipping of the right distal anterior cerebral artery mycotic aneurysm. (a) Frontal view of digital subtraction angiography (DSA) of the right internal carotid artery shows a saccular distal right M1 aneurysm (arrowhead). (b) Volumetric rendering depicts the saccular aneurysm (arrowhead), allowing accurate preoperative measurement of its size, height, and neck. The arterial supply from the right middle cerebral artery (MCA) and its angulation in three-dimensional space are also visualised. (c) DSA of the right MCA using the roadmap technique enabled neuronavigation with the use of a guidewire to access the M1 aneurysm for precise coil embolisation (arrows). (d) Post-embolisation DSA of the right MCA shows successful occlusion of the mycotic aneurysm with preserved flow to the distal branches (arrow). (e) Frontal DSA of the right vertebral artery demonstrates a peripherally located P4 aneurysm (arrowhead). (f) Volumetric rendering depicts its saccular morphology with a narrow neck and clearly shows the arterial supply (arrowhead) from the left posterior cerebral artery (PCA), aiding in accurate preoperative planning. (g) DSA of the left PCA enabled direct neuronavigation using a guidewire (arrows) to the target aneurysm for embolisation. (h) DSA of the distal left PCA shows precise coil embolisation of the peripherally located mycotic aneurysm (arrow), performed through an indirect approach that resulted in parent artery occlusion. Despite the challenging locations of the mycotic aneurysms, DSA with volumetric rendering and the roadmap technique allowed successful neuronavigated embolisation.

 

Figure 8. Same patient as Figures 3 and 4. Follow-up computed tomography angiography (CTA) 2 months later showed a persistent M2 mycotic aneurysm (not shown). (a) Oblique projection of digital subtraction angiography (DSA) of the right internal carotid artery (ICA) shows a lobulated M2 mycotic aneurysm arising from the middle cerebral artery (arrowhead). (b) DSA with the roadmap technique enabled neuronavigation for precise embolisation of the mycotic aneurysm (arrow). (c) Post-embolisation DSA of the right ICA shows the successfully embolised right M2 aneurysm (arrow), with preservation of distal flow. (d) Follow-up coronal cone beam CTA 2 months post-embolisation shows a new saccular aneurysm adjacent to the previously embolised aneurysm (arrow). (e) Oblique projection DSA of the right ICA confirms the presence of the new narrow-neck mycotic aneurysm arising from the medial wall of the M2 segment (arrow). (f) Coil embolisation of the second aneurysm was performed, and a flow diverter was deployed across the aneurysmal neck. Follow-up sagittal cone beam CTA 1.5 years later shows the flow diverter in situ (arrow), with successfully embolised aneurysm (arrowhead). The stent remains patent and the distal branches are preserved.

 

Neurological complications secondary to IE require prompt recognition of its typical presentations and imaging manifestations to facilitate early diagnosis of neurological complications and their subsequent treatment, including possible radiological intervention.