SA Journal of Radiology最新文献

Background: Hepatocellular carcinoma (HCC) is highly prevalent in sub-Saharan Africa (SSA). LI-RADS^® is a standardised system for imaging-based diagnosis and characterisation of HCC.

Objectives: This study assessed knowledge, attitudes and current practices related to LI-RADS in SSA, with a view to identifying barriers to its utilisation and informing targeted educational interventions.

Method: A 21-item anonymous electronic questionnaire was distributed to medical professionals in SSA using the SurveyMonkey online platform. Knowledge, attitudes and current practices regarding LI-RADS^® were assessed. Data were analysed using descriptive statistics, and comparisons were made between radiologists and non-radiologists.

Results: There were 134 respondents from 14 of the 34 SSA countries. Radiologists significantly outperformed non-radiologists in LI-RADS^® knowledge, particularly regarding its purpose (65.6% vs 38.2%, p = 0.0007), arterial phase hyperenhancement definition (90.3% vs 59.8%, p = 0.004) and size criteria (77.4% vs 45.1%, p = 0.003) for diagnosing hepatocellular carcinoma (HCC). However, 43.8% of radiologists and 63.4% of non-radiologists did not recognise the limitations of LI-RADS^®. Only 34.3% stated that LI-RADS^® was their reporting standard and 29.1% of the respondents indicated that less than 25% of their radiological reports adhered to LI-RADS^®. The majority (78.3%) of participants stated they preferred radiology reports for high-risk liver lesions to be LI-RADS^®-standardised. The two main barriers to adopting LI-RADS^® included lack of consistency (44.8%) and unfamiliarity with the reporting system (27.6%).

Conclusion: Although LI-RADS^® remains the preferred reporting system for HCC, there are significant gaps in its knowledge and implementation across SSA.

Contribution: This survey highlights the needs for targeted educational initiatives and improved training to enhance the adoption and use of LI-RADS^® in SSA.

Background: Many South African peripheral medical centres lack direct access to CT scans or neurosurgery. The Glasgow Coma Scale (GCS), used with or without other findings, remains widely utilised in traumatic brain injury (TBI) assessments with lack of standardisation between centres. There is limited data from South Africa (SA) correlating GCS scores to CT imaging in TBI.

Objectives: This study aimed to assess CT findings at various GCS levels to determine whether GCS was a reliable indicator for imaging and referral.

Method: A retrospective review of 385 patients categorised with mild, moderate or severe TBI was performed. The initial non-sedated post-resuscitation GCS score and initial CT brain findings were compared using the chi-square and Fisher's exact tests.

Results: Increased intracranial pressure and subdural haemorrhage occurred in 41.7% and 53.7% of patients with GCS 9-12, respectively, and 30.5% and 41.4% of patients with GCS 13-15, respectively. The highest incidence of depressed skull fractures (51.3%; 95% confidence interval [CI], 43.2-59.3%; p < 0.001) and pneumocephalus (25.6%; 95% CI, 42.2-56.4; p < 0.001) were reported in the CGS 13-15 category. Neurosurgical intervention was required in 83.2% and 73.0% of patients with GCS scores of 9-12 and 13-15, respectively.

Conclusion: The severe category of GCS predicts imaging and neurosurgery requirements while the mild to moderate categories underpredict the need for patient referral.

Contribution: This study provides rationale for the development of a local, standardised assessment tool to guide referral of TBI patients for imaging in resource-limited settings.

Conductive hearing loss (CHL) results from impaired mechanical transmission of sound through the external or middle ear and is commonly associated with conditions such as otosclerosis, cholesteatoma, ossicular discontinuity, congenital malformations and trauma. Less frequently, it may arise from vascular or neoplastic lesions. Imaging plays a central role in the evaluation of CHL, with high-resolution CT (HRCT) regarded as the gold standard for initial assessment because of its superior spatial resolution in detecting bony pathologies, including fenestral otosclerosis, ossicular chain defects, third-window lesions and trauma-related disruptions. MRI complements HRCT by providing excellent soft tissue contrast, facilitating the detection of postoperative cholesteatoma via non-echo-planar diffusion-weighted imaging and the characterisation of vascular tumours such as glomus tympanicum. CT angiography (CTA), while not routinely indicated, can be valuable for identifying vascular anomalies and aiding surgical planning in selected cases. Technical considerations, such as optimised scan parameters, strategies for radiation dose reduction in paediatric patients and the integration of multiple imaging modalities, are essential for accurate diagnosis and effective treatment planning. Collectively, HRCT, MRI and CTA provide a structured, evidence-based framework for the comprehensive evaluation and management of CHL.

Contribution: This review synthesises current evidence on HRCT, MRI and CTA in the assessment of CHL, emphasising their complementary roles, protocol optimisation and multimodal integration to enhance diagnostic accuracy and surgical guidance in both paediatric and adult populations.

This case report highlights the findings and treatment in a patient with a splenic arteriovenous fistula causing non-cirrhotic portal hypertension (NCPH). Splenic arteriovenous malformations (AVMs) are rare, and their management using percutaneous embolisation is an emerging alternative modality as opposed to open surgical intervention.

Contribution: Similar cases have been published; however, few highlight the management role of interventional radiology.

Congenital depressed skull fractures (ping-pong fractures) without obstetric trauma are rare. A term male neonate delivered via uncomplicated caesarean section, demonstrated a right parieto-temporal skull depression (5 cm × 5 cm) at birth. Computed tomography revealed a 4 mm parietal depression without intracranial injury. No instrumental delivery or maternal trauma were present. The likely aetiology was intrauterine compression ('faulty foetal packing'). The patient was managed conservatively with close follow-up.

Contribution: This case underscores the importance of perinatal history and neuroimaging in distinguishing spontaneous from traumatic fractures and supports conservative management in neurologically intact infants.

The mediastinum is an unusual location for cystic lymphangiomas. An 18-month-old male presented with acute fever, dry intermittent cough and respiratory distress. Chest radiography, ultrasonography, contrast-enhanced CT and MRI suggested a diagnosis of mediastinal cystic lymphangioma with internal haemorrhage. Surgical excision of the lesion and histopathological examination confirmed cystic lymphangioma.

Contribution: This case highlights the multimodal radiological features of isolated mediastinal cystic lymphangioma for accurate diagnosis and improved management, to avoid unnecessary interventions and complications.

Background: Assessment of spinal trauma entails a full neurological examination and radiological assessment to determine the level of spinal cord injury.

Objectives: This study aimed to determine if further imaging is always required, whether the clinical picture correlates with imaging results and to compare clinical and radiological prediction accuracy.

Method: This retrospective chart review compared and correlated clinical findings with radiological findings in patients with spinal trauma at Inkosi Albert Luthuli Central Hospital over a period of 6 years. Demographics and sensitivity and specificity of clinical to imaging correlation with positive predictive ratios were assessed.

Results: A total of 290 patients admitted with spinal injury, who received CT and/or MRI, were evaluated. Cervical-spine injuries were common. For predicting abnormal CT findings, the sensitivity of motor and sensory findings was 69.2% with a specificity of 85.4%. The positive predictive value (PPV) of motor and sensory findings was 96.2%. The negative predictive value (NPV) of motor and sensory findings was 34.0%. On MRI, sensitivity for motor and sensory findings was 85.1% for correctly predicting abnormal MRI findings, while the specificity was 52.8%. The PPV of motor and sensory findings was 82.5% with a NPV of 57.6%.

Conclusion: In this trauma population, correlation of clinical findings with abnormal CT findings was 84.4% and for MRI findings was 72.3%, indicating that clinical findings alone may not be sufficient to rule out the need for imaging; false negatives could lead to missed or incorrect level of injury diagnoses.

Contribution: This study adds to the proof that while clinical findings are reasonably accurate for the determination of neurological spinal cord injury level, both CT and MRI add additional information, making these tests invaluable.

Background: Triple-negative breast cancer (TNBC) is considered an aggressive subtype, defined by the absence of oestrogen, progesterone and HER2 receptors. It typically presents earlier and more aggressively. Limited data exist on its prevalence, age of onset and imaging features in South Africa.

Objectives: This study aimed to assess the prevalence of TNBC at Helen Joseph Tertiary Hospital (HJTH), describe its histopathological features and explore trends in age at diagnosis and imaging patterns-including early-onset disease.

Method: A retrospective review of 280 female patients with histologically confirmed breast cancer, diagnosed between January 2021 and December 2023, was conducted. Demographic, imaging and histopathology data were analysed using descriptive statistics and chi-square tests.

Results: The diagnosis of TNBC accounted for 17% (48/280) of all breast cancer cases in the cohort. The TNBC lesions typically measured 1-5 cm and showed nodal involvement in 73% of cases. Despite their aggressive biology, many TNBC lesions appeared circumscribed or only mildly irregular on imaging, mimicking benign masses. Among all the 280 breast cancer cases, 61% were high-grade. The mean Ki-67 index for TNBC was the highest at 52%, followed by HER2+ (39%), Luminal B (33%) and Luminal A (21%). Notably, some HER2+ and TNBC cases exhibited lower Ki-67 indices, highlighting heterogeneity within these subtypes.

Conclusion: This study highlights the complexity of breast cancer presentation in a South African setting, particularly the discordance between tumour biology and imaging.

Contribution: These findings contribute local data on TNBC in an urban public healthcare context, supporting improved imaging awareness and clinical vigilance in resource-limited settings.

Background: Interventional radiology (IR) is rapidly growing as a critical subspeciality, yet in South Africa (SA) IR training remains underdeveloped, with no national data to guide structured training and workforce expansion.

Objectives: This study aimed to assess South African radiology registrars' interest in IR, their exposure, and the existing training environment.

Method: A cross-sectional, electronic quantitative and qualitative survey was conducted from 09 February 2024 to 08 March 2024 among registrars affiliated with the Radiological Society of South Africa. The survey gauged their interest in IR, exposure, training infrastructure, learning tools, satisfaction with training and potential areas for improvement.

Results: Response rate: 45% (100/221 registrars); 54% were female, median age was 33.4 years and 51% in their first or second year. Interest in IR as a subspeciality was expressed by 40%, 27% were not interested and 33% were unsure. Only 27% reported adequate exposure; 56% had limited exposure and 17% had none. Dissatisfaction with IR training was noticed by 47%, 37% were undecided and 16% were satisfied. Only 9% had a structured curriculum, while 23% relied on mentorship. Major themes for improvement included structured training, mentorship, supervision, training facilities and hands-on exposure.

Conclusion: Despite strong interest, radiology registrars (including international supernumerary registrars) face significant gaps in IR training and exposure in SA.

Contribution: This is the first national study to provide essential baseline data to inform targeted reforms and development of structured, locally relevant IR training.

Background: Breast cancer remains a major public health burden in South Africa, with diagnostic delays contributing to poor outcomes. Ultrasound is effective for early detection but is limited by access and operator variability. Integrating artificial intelligence (AI) into point-of-care ultrasound (POCUS) offers a potential solution.

Objectives: To evaluate the diagnostic performance of a locally developed AI-enabled POCUS system (Breast AI) in predicting malignancy among women with palpable breast abnormalities.

Method: A prospective cohort study was conducted between June 2024 and November 2024 at Groote Schuur Hospital. Women aged ≥ 25 years with suspicious breast lesions underwent Breast AI ultrasound prior to biopsy. Real-time malignancy risk scores were compared with histopathological results. Diagnostic accuracy was assessed using sensitivity, specificity, positive predictive value (PPV), F1 score and area under the curve (AUC).

Results: Among 159 participants, Breast AI achieved a sensitivity of 67.2%, specificity of 79.4% and PPV of 70.3% at a 51% threshold. The AUC was 0.76, reflecting moderate discriminatory performance. F1 score analysis identified 51% as the optimal cut-off (F1 = 65.7%). Benign pathologies such as fibroadenomas and fat necrosis correlated with low AI scores. A three-tiered risk model was developed: < 30% (low), 30% - 51% (intermediate) and > 51% (high risk).

Conclusion: Breast AI demonstrates promising diagnostic accuracy for triaging suspicious breast lesions, particularly in resource-constrained settings.

Contribution: This study provides real-world evidence supporting the integration of AI into POCUS to improve breast cancer detection and clinical decision-making in low-resource environments.