Radiology
Audio Summaries

From ACR O-RADS 2022 to Explainable Deep Learning: Comparative Performance of Expert Radiologists, Convolutional Neural Networks, Vision Transformers, and Fusion Models for Ultrasound-Based Risk Stratification of Ovarian Masses

This study aims to evaluate the performance of expert radiologists using the updated Ovarian-Adnexal Reporting and Data System (O-RADS) v2022 for risk stratification of ovarian masses and to compare it with various deep learning (DL) models, including convolutional neural networks (CNNs) and vision transformers (ViTs). The findings indicate that while DL models, particularly ViT16-384, outperform radiologist assessments, the integration of radiologist scores with AI significantly enhances diagnostic accuracy, suggesting that hybrid human-AI frameworks could standardize ultrasound interpretation and improve the identification of high-risk ovarian lesions.

High-Frequency Ultrasound Assessment of Female Genital Rejuvenation

The authors aimed to evaluate the role of high-frequency ultrasound (HFUS) in assessing and monitoring female genital rejuvenation procedures. Their findings indicate that HFUS effectively visualizes anatomical structures and treatment-related changes, revealing distinct sonographic patterns associated with various aesthetic treatments. These results suggest that HFUS could be a valuable tool in aesthetic practice, warranting further research for standardization.

Real or not real? Can radiologists distinguish artificial intelligence generated radiological images from real ones?

The authors aimed to assess whether radiologists can distinguish between AI-generated and real radiological images, as well as identify factors influencing their classification accuracy. The study found that radiologists correctly identified about 77.8% of the images, with performance varying by image type and being enhanced by relevant specialist expertise, but not influenced by years of experience or familiarity with AI.

Time-dependent diffusion MRI for assessing tumor microstructure and prognostic risk factors in cervical cancer

The authors aimed to assess the utility of time-dependent diffusion MRI (td-dMRI) in noninvasively characterizing tumor microstructure and predicting prognostic risk factors in cervical cancer. Their findings indicate that specific td-dMRI parameters, particularly the intracellular water exchange time and certain diffusion coefficients, may serve as potential imaging biomarkers for differentiating tumor types and assessing histological grade and lymphovascular space invasion. However, the authors emphasize the need for further validation in larger, multicenter studies.

A clinical-radiomics model based on MRI sub-regions of gluteus maximus for recurrence prediction in high-grade serous ovarian cancer

The authors aimed to evaluate the predictive value of MRI sub-regional radiomics of the gluteus maximus for recurrence in high-grade serous ovarian cancer (HGSOC) and to compare its performance with conventional radiomics and deep learning models. Their findings indicate that the sub-regional radiomics model outperformed traditional methods, achieving higher predictive accuracy when combined with clinical factors, thus offering a valuable tool for assessing recurrence risk in HGSOC patients post-treatment.

Diagnostic yield of 8G versus 11G transpedicular vertebral biopsy needles in suspected tuberculous spondylitis: a prospective randomized controlled study

The authors aimed to compare the diagnostic yield and complication rates of 8G versus 11G transpedicular vertebral biopsy needles in patients suspected of tuberculous spondylitis. The study found no significant differences in diagnostic outcomes or complications between the two needle sizes, with both groups yielding similar rates of diagnostic samples and complications. Overall, the results suggest that either needle size can be effectively used for biopsies in this context.

Autonomous AI in prostate cancer: the road ahead towards clinical implementation

The authors investigate the barriers to the clinical implementation of autonomous AI for detecting clinically significant prostate cancer (csPCa) on MRI, despite its promising diagnostic performance. They identify key challenges in evidence, safety, and acceptance among patients and radiologists, emphasizing the need for large-scale trials, improved transparency, and education to facilitate the responsible deployment of this technology in clinical practice.

Autoimmune diseases involving the gastrointestinal tract: clinical patterns, imaging spectrum, and complications

The authors aim to elucidate the clinical patterns, imaging characteristics, and complications associated with autoimmune diseases that affect the gastrointestinal tract. They categorize these diseases into three groups based on their primary or secondary involvement of the GI system and emphasize the importance of integrating imaging findings with clinical and serologic data to improve diagnosis and management. This review highlights how advances in imaging can enhance understanding and recognition of these complex conditions.

Local recurrence in rectal cancer: from detection to structured reporting

The authors investigate the challenges and advancements in detecting local recurrence of rectal cancer following total mesorectal excision and neoadjuvant therapy, which affects a significant percentage of patients. They emphasize the importance of early recognition through multimodal imaging techniques, particularly MRI, to differentiate between viable tumors and postoperative changes, and discuss the role of structured reporting and emerging technologies in improving patient management. The review aims to provide a comprehensive overview of recurrence patterns and optimal imaging strategies to enhance surgical planning and outcomes.

Scaling AI-enabled imaging-based screening: lessons from reporting for the NHS England lung cancer screening programme

The authors aim to explore the scalability and transferability of the NHS England Lung Cancer Screening Programme's infrastructure for imaging-based screening. They highlight three key components: a cloud-native imaging IT system, a national network of specialized radiologists, and the integration of AI for nodule detection, which collectively enhance the efficiency and effectiveness of lung cancer screening. The findings suggest that these principles could be adapted for other screening modalities, such as prostate MRI and CT colonography, indicating a broader applicability of their approach.