Patient-centered radiology reports with generative artificial intelligence: adding value to radiology reporting.

The purposes were to assess the efficacy of AI-generated radiology reports in terms of report summary, patient-friendliness, and recommendations and to evaluate the consistent performance of report quality and accuracy, contributing to the advancement of radiology workflow. Total 685 spine MRI reports were retrieved from our hospital database. AI-generated radiology reports were generated in three formats: (1) summary reports, (2) patient-friendly reports, and (3) recommendations. The occurrence of artificial hallucinations was evaluated in the AI-generated reports. Two radiologists conducted qualitative and quantitative assessments considering the original report as a standard reference. Two non-physician raters assessed their understanding of the content of original and patient-friendly reports using a 5-point Likert scale. The scoring of the AI-generated radiology reports were overall high average scores across all three formats. The average comprehension score for the original report was 2.71 ± 0.73, while the score for the patient-friendly reports significantly increased to 4.69 ± 0.48 (p < 0.001). There were 1.12% artificial hallucinations and 7.40% potentially harmful translations. In conclusion, the potential benefits of using generative AI assistants to generate these reports include improved report quality, greater efficiency in radiology workflow for producing summaries, patient-centered reports, and recommendations, and a move toward patient-centered radiology.

Sarcopenia prediction using shear-wave elastography, grayscale ultrasonography, and clinical information with machine learning fusion techniques: feature-level fusion vs. score-level fusion.

This study aimed to develop and evaluate a sarcopenia prediction model by fusing numerical features from shear-wave elastography (SWE) and gray-scale ultrasonography (GSU) examinations, using the rectus femoris muscle (RF) and categorical/numerical features related to clinical information. Both cohorts (development, 70 healthy subjects; evaluation, 81 patients) underwent ultrasonography (SWE and GSU) and computed tomography. Sarcopenia was determined using skeletal muscle index calculated from the computed tomography. Clinical and ultrasonography measurements were used to predict sarcopenia based on a linear regression model with the least absolute shrinkage and selection operator (LASSO) regularization. Furthermore, clinical and ultrasonography features were combined at the feature and score levels to improve sarcopenia prediction performance. The accuracies of LASSO were 70.57 ± 5.00-81.54 ± 4.83 (clinical) and 69.00 ± 4.52-69.73 ± 5.47 (ultrasonography). Feature-level fusion of clinical and ultrasonography (accuracy, 70.29 ± 6.63 and 83.55 ± 4.32) showed similar performance with clinical features. Score-level fusion by AdaBoost showed the best performance (accuracy, 73.43 ± 6.57-83.17 ± 5.51) in the development and evaluation cohorts, respectively. This study might suggest the potential of machine learning fusion techniques to enhance the accuracy of sarcopenia prediction models and improve clinical decision-making in patients with sarcopenia.

Clinical Applications, Challenges, and Recommendations for Artificial Intelligence in Musculoskeletal and Soft-Tissue Ultrasound: AJR Expert Panel Narrative Review.

Artificial intelligence (AI) is increasingly used in clinical practice for musculoskeletal imaging tasks, such as disease diagnosis and image reconstruction. AI applications in musculoskeletal imaging have focused primarily on radiography, CT, and MRI. Although musculoskeletal ultrasound stands to benefit from AI in similar ways, such applications have been relatively underdeveloped. In comparison with other modalities, ultrasound has unique advantages and disadvantages that must be considered in AI algorithm development and clinical translation. Challenges in developing AI for musculoskeletal ultrasound involve both clinical aspects of image acquisition and practical limitations in image processing and annotation. Solutions from other radiology subspecialties (e.g., crowdsourced annotations coordinated by professional societies), along with use cases (most commonly rotator cuff tendon tears and palpable soft-tissue masses), can be applied to musculoskeletal ultrasound to help develop AI. To facilitate creation of high-quality imaging datasets for AI model development, technologists and radiologists should focus on increasing uniformity in musculoskeletal ultrasound performance and increasing annotations of images for specific anatomic regions. This Expert Panel Narrative Review summarizes available evidence regarding AI's potential utility in musculoskeletal ultrasound and challenges facing its development. Recommendations for future AI advancement and clinical translation in musculoskeletal ultrasound are discussed.

3-D breast nodule detection on automated breast ultrasound using faster region-based convolutional neural networks and U-Net.

Mammography is currently the most commonly used modality for breast cancer screening. However, its sensitivity is relatively low in women with dense breasts. Dense breast tissues show a relatively high rate of interval cancers and are at high risk for developing breast cancer. As a supplemental screening tool, ultrasonography is a widely adopted imaging modality to standard mammography, especially for dense breasts. Lately, automated breast ultrasound imaging has gained attention due to its advantages over hand-held ultrasound imaging. However, automated breast ultrasound imaging requires considerable time and effort for reading because of the lengthy data. Hence, developing a computer-aided nodule detection system for automated breast ultrasound is invaluable and impactful practically. This study proposes a three-dimensional breast nodule detection system based on a simple two-dimensional deep-learning model exploiting automated breast ultrasound. Additionally, we provide several postprocessing steps to reduce false positives. In our experiments using the in-house automated breast ultrasound datasets, a sensitivity of [Formula: see text] with 8.6 false positives is achieved on unseen test data at best.

Early detection of diabetic retinopathy based on deep learning and ultra-wide-field fundus images.

Visually impaired and blind people due to diabetic retinopathy were 2.6 million in 2015 and estimated to be 3.2 million in 2020 globally. Though the incidence of diabetic retinopathy is expected to decrease for high-income countries, detection and treatment of it in the early stages are crucial for low-income and middle-income countries. Due to the recent advancement of deep learning technologies, researchers showed that automated screening and grading of diabetic retinopathy are efficient in saving time and workforce. However, most automatic systems utilize conventional fundus photography, despite ultra-wide-field fundus photography provides up to 82% of the retinal surface. In this study, we present a diabetic retinopathy detection system based on ultra-wide-field fundus photography and deep learning. In experiments, we show that the use of early treatment diabetic retinopathy study 7-standard field image extracted from ultra-wide-field fundus photography outperforms that of the optic disc and macula centered image in a statistical sense.