Radiological Differential Diagnoses Based on Cardiovascular and Thoracic Imaging Patterns: Perspectives of Four Large Language Models.

Background Differential diagnosis in radiology is a critical aspect of clinical decision-making. Radiologists in the early stages may find difficulties in listing the differential diagnosis from image patterns. In this context, the emergence of large language models (LLMs) has introduced new opportunities as these models have the capacity to access and contextualize extensive information from text-based input. Objective The objective of this study was to explore the utility of four LLMs-ChatGPT3.5, Google Bard, Microsoft Bing, and Perplexity-in providing most important differential diagnoses of cardiovascular and thoracic imaging patterns. Methods We selected 15 unique cardiovascular ( n = 5) and thoracic ( n = 10) imaging patterns. We asked each model to generate top 5 most important differential diagnoses for every pattern. Concurrently, a panel of two cardiothoracic radiologists independently identified top 5 differentials for each case and came to consensus when discrepancies occurred. We checked the concordance and acceptance of LLM-generated differentials with the consensus differential diagnosis. Categorical variables were compared by binomial, chi-squared, or Fisher's exact test. Results A total of 15 cases with five differentials generated a total of 75 items to analyze. The highest level of concordance was observed for diagnoses provided by Perplexity (66.67%), followed by ChatGPT (65.33%) and Bing (62.67%). The lowest score was for Bard with 45.33% of concordance with expert consensus. The acceptance rate was highest for Perplexity (90.67%), followed by Bing (89.33%) and ChatGPT (85.33%). The lowest acceptance rate was for Bard (69.33%). Conclusion Four LLMs-ChatGPT3.5, Google Bard, Microsoft Bing, and Perplexity-generated differential diagnoses had high level of acceptance but relatively lower concordance. There were significant differences in acceptance and concordance among the LLMs. Hence, it is important to carefully select the suitable model for usage in patient care or in medical education.

Assessing the Capability of ChatGPT, Google Bard, and Microsoft Bing in Solving Radiology Case Vignettes.

Background The field of radiology relies on accurate interpretation of medical images for effective diagnosis and patient care. Recent advancements in artificial intelligence (AI) and natural language processing have sparked interest in exploring the potential of AI models in assisting radiologists. However, limited research has been conducted to assess the performance of AI models in radiology case interpretation, particularly in comparison to human experts. Objective This study aimed to evaluate the performance of ChatGPT, Google Bard, and Bing in solving radiology case vignettes (Fellowship of the Royal College of Radiologists 2A [FRCR2A] examination style questions) by comparing their responses to those provided by two radiology residents. Methods A total of 120 multiple-choice questions based on radiology case vignettes were formulated according to the pattern of FRCR2A examination. The questions were presented to ChatGPT, Google Bard, and Bing. Two residents wrote the examination with the same questions in 3 hours. The responses generated by the AI models were collected and compared to the answer keys and explanation of the answers was rated by the two radiologists. A cutoff of 60% was set as the passing score. Results The two residents (63.33 and 57.5%) outperformed the three AI models: Bard (44.17%), Bing (53.33%), and ChatGPT (45%), but only one resident passed the examination. The response patterns among the five respondents were significantly different ( p = 0.0117). In addition, the agreement among the generative AI models was significant (intraclass correlation coefficient [ICC] = 0.628), but there was no agreement between the residents (Kappa = -0.376). The explanation of generative AI models in support of answer was 44.72% accurate. Conclusion Humans exhibited superior accuracy compared to the AI models, showcasing a stronger comprehension of the subject matter. All three AI models included in the study could not achieve the minimum percentage needed to pass an FRCR2A examination. However, generative AI models showed significant agreement in their answers where the residents exhibited low agreement, highlighting a lack of consistency in their responses.

Postoperative Delayed Facial Nerve Palsy: A Surgeon's Nightmare.

The facial nerve supplies motor, sensory, and parasympathetic innervation to the head and neck, and its paralysis can have significant physical and psychological impacts. This study discusses a compelling case involving a 21-year-old male who developed delayed facial nerve palsy (DFNP) on the eighth day after cortical mastoid surgery. Through conservative management, the patient achieved a full recovery by the 52nd day. Our experience underscores the importance of approaching DFNP with patience, emphasizing the need for thorough counseling of both the patient and their family members.

Analysing lumbar pedicle morphometry observed via traditional and recent modalities.

Objective: The present study was conducted to collect morphometric data on the lumbar vertebrae pedicles of the adult population from the eastern parts of India and analyse the variations, if any, with other parts of the country and the world. Methods: The retrospective cross-sectional study where lumbar pedicle morphometric data was obtained via dried bone, the 3D Lumbar vertebrae images were obtained by scanning the dried lumbar vertebrae, and the 3D lumbar vertebrae model was generated from a 1 mm thin CT scan slice of the Lumbar spine of patients who were advised to have a CT scan of the abdomen for reasons other than related to the vertebral column. Both linear and angular measurements in the lumbar pedicles were made bilaterally. Results: The transverse pedicle width is widest at L3 and the narrowest at L2 vertebra bilaterally. The sagittal pedicle width at L4 vertebrae was observed to be the widest bilaterally, while L3 had the narrowest pedicle. The pedicular and body length along the pedicular axis is longer than the central axis at all the lumbar levels. The linear measurement along the pedicular axis was longest at L5 bilaterally via both modes, with a range of 18.2-47.31 mm for bones and 21.03-49.28 mm for CT scan morphometry. The transverse pedicular angle on analysis was observed to increase as one goes down the spine from L1 to L5, with a steep rise between L4 and L5. In contrast, the sagittal pedicular angle decreased as we went caudally toward the L5 vertebrae. Conclusion: The present study data had significant differences among the values reported in the literature from the different populations for the parameters studied. The data obtained by this study will be highly beneficial for the success of the free-hand technique of pedicle screw insertion.

Subspecialization, Senior Residency, or Private Practice: The Dilemma of Final-Year Radiology Postgraduate Residents in India.

Background Radiology is among India's top five speciality choices pursued by meritorious medical graduates. With the advent of the subspecialization degree courses, fellowships, the requirement of senior residency as faculty eligibility criteria, and the lucrative option of private practice, the final-year postgraduates are given opportunities to choose from but with little guidance on what and how to choose. This study aims to analyze the views of the final-year radiology postgraduate residents in India regarding their understanding of how to proceed in their professional life with options available. Materials and Methods A cross-sectional, questionnaire-based study was conducted online via Google forms distributed via email and social media platforms. The questionnaire was prepared after going through previous literature, video blogs, and media available on the Internet and was further validated for content. Results About half (48%) of them wished to pursue higher studies in the form of Doctorate of Medicine (DM) degree courses or fellowships. Almost three-quarters of the participants preferred short-term subspecialization via fellowships over 3-year DM courses (28%). Regarding clinical practice, 61.9% preferred to take up senior residency, while slightly over one-third (35.7%) expressed their will to move on to private practice. Conclusion A relative conundrum was observed in the decision to take up senior residency or private practice or to go for DM but had to pursue a fellowship due to limited choice in topics and seats available in the country for subspecialization.

Assessing ChatGPT's Proficiency in Simplifying Radiological Reports for Healthcare Professionals and Patients.

Background Clear communication of radiological findings is crucial for effective healthcare decision-making. However, radiological reports are often complex with technical terminology, making them challenging for non-radiology healthcare professionals and patients to comprehend. Large language models like ChatGPT (Chat Generative Pre-trained Transformer, by OpenAI, San Francisco, CA) offer a potential solution by translating intricate reports into simplified language. This study aimed to assess the capability of ChatGPT-3.5 in simplifying radiological reports to facilitate improved understanding by healthcare professionals and patients. Materials and methods Nine radiological reports were taken for this study spanning various imaging modalities and medical conditions. These reports were used to ask ChatGPT a set of seven questions (describe the procedure, mention the key findings, express in a simple language, suggestions for further investigation, need of further investigation, grammatical or typing errors, and translation into Hindi). A total of eight radiologists rated the generated content in detailing, summarizing, simplifying content and language, factual correctness, further investigation, grammatical errors, and translation to Hindi. Results The highest score was obtained for detailing the report (94.17% accuracy) and the lowest score was for drawing conclusions for the patient (85% accuracy); case-wise scores were similar (p-value = 0.97). The Hindi translation by ChatGPT was not suitable for patient communication. Conclusion The current free version of ChatGPT-3.5 was able to simplify radiological reports effectively, removing technical jargon while preserving essential diagnostic information. The free version adeptly simplifies radiological reports, enhancing accessibility for healthcare professionals and patients. Hence, it has the potential to enhance medical communication, facilitating informed decision-making by healthcare professionals and patients.

Glutaric Aciduria Type I: A Rare Metabolic Disorder Mimicking as Choreoathetoid Cerebral Palsy.

Glutaric aciduria type I (GA I) is an autosomal recessive inborn error of metabolism caused by a deficiency of the enzyme glutaryl-CoA dehydrogenase. This disorder is characterized by progressive dystonia, choreoathetosis, and dyskinesia. It is often misdiagnosed as athetoid cerebral palsy. Laboratory evaluation usually demonstrates increased urinary excretion of gluataric acid and 3-hydroxyglutaric acid. We report a case of a 7-year-old boy presenting with choreoathetosis and dystonia, mimicking as choreoathetoid cerebral palsy. The presence of characteristic neuroimaging and biochemical studies led to the diagnosis of GA I.

Combined Left Recurrent Laryngeal Nerve and Phrenic Nerve Palsy: A Rare Presentation of Thoracic Aortic Aneurysm.

Hoarseness of voice due to Vocal Cord Palsy (VCP) resulting from aortic aneurysm is a rare entity. Hoarseness due to left Recurrent Laryngeal Nerve (RLN) paralysis caused by identifiable cardiovascular disease has been described as Ortner's syndrome or Cardiovocal syndrome. Very rarely, thoracic aortic aneurysm can cause Phrenic Nerve (PN) palsy causing hemidiaphragm paralysis. But, aortic aneurysm causing both RLN and PN palsy is an extremely rare occasion. Our literature review showed only three cases of combined RLN and PN palsy due to thoracic aortic aneurysm. Here we report a 70-year-old patient having hoarseness of voice for six years for which he was on symptomatic treatment. He presented to us with cough and blood tinged sputum, left sided chest pain, loss of appetite and weakness since two months duration. Based on clinical history and chest radiograph, lung malignancy was suspected. To our surprise, Contrast Enhanced Computed Tomography (CECT) revealed partially thrombosed saccular thoracic aneurysm originating distal to origin of left subclavian artery. He was referred to the cardiothoracic vascular surgeon for surgical treatment, but was reluctant to go for surgery. He is now being followed up by cardiothoracic vascular surgery and cardiology department and his condition has remained the same.

Leiomyosarcoma of the prostate-an unexpected histopathological outcome.

Prostate leiomyosarcoma is an extremely rare and highly aggressive neoplasm that accounts for >0.1% of all primary prostate malignancies. We report a case of a patient, presenting with recurrent episodes of dysuria, who had been diagnosed and operated for benign prostatic hyperplasia 1 month earlier, and now presented with similar symptoms postoperatively. Trans-rectal biopsy of the prostate was carried out and histopathology revealed leiomyosarcoma of the prostate.