Keeping Up With ChatGPT: Evaluating Its Recognition and Interpretation of Nuclear Medicine Images.

PURPOSE: The latest iteration of GPT4 (generative pretrained transformer) is a large multimodal model that can integrate both text and image input, but its performance with medical images has not been systematically evaluated. We studied whether ChatGPT with GPT-4V(ision) can recognize images from common nuclear medicine examinations and interpret them. PATIENTS AND METHODS: Fifteen representative images (scintigraphy, 11; PET, 4) were submitted to ChatGPT with GPT-4V(ision), both in its Default and "Advanced Data Analysis (beta)" version. ChatGPT was asked to name the type of examination and tracer, explain the findings and whether there are abnormalities. ChatGPT should also mark anatomical structures or pathological findings. The appropriateness of the responses was rated by 3 nuclear medicine physicians. RESULTS: The Default version identified the examination and the tracer correctly in the majority of the 15 cases (60% or 53%) and gave an "appropriate" description of the findings or abnormalities in 47% or 33% of cases, respectively. The Default version cannot manipulate images. "Advanced Data Analysis (beta)" failed in all tasks in >90% of cases. A "major" or "incompatible" inconsistency between 3 trials of the same prompt was observed in 73% (Default version) or 87% of cases ("Advanced Data Analysis (beta)" version). CONCLUSIONS: Although GPT-4V(ision) demonstrates preliminary capabilities in analyzing nuclear medicine images, it exhibits significant limitations, particularly in its reliability (ie, correctness, predictability, and consistency).

Reevaluating the relevance of 18F-FDG PET findings for diagnosis of neurosarcoidosis: a case series.

OBJECTIVE: The diagnosis of neurosarcoidosis (NS) remains challenging due to the difficulty to obtain central nervous system (CNS) biopsies. Various diagnostic parameters are considered for the definition of possible, probable and definite NS. Magnetic resonance imaging (MRI) is the imaging gold standard and considered in diagnostic criteria. Fluorodeoxyglucose positron emission (18F-FDG PET) is sometimes performed additionally to identify possible systemic biopsy targets. However, at present, its findings are not incorporated into the diagnostic criteria for neurosarcoidosis (NS). METHODS: We conducted a single center retrospective search for the period 2020-2022, for patients with neurological symptoms in a diagnostic context of suspected NS who underwent MRI and additional 18F-FDG PET scans to identify potential hypermetabolism in the CNS and biopsy targets. RESULTS: We identified three cases of NS, where Gadolinium-enhanced MRI scans did not show abnormalities while 18F-FDG PET revealed hypermetabolic lesions in areas of the CNS. Additional MRI scans were still inconclusive for structural changes. We diagnosed a "probable" NS in all cases with histopathological confirmation of systemic sarcoidosis which led to an intensified therapy regime. DISCUSSION: 18F-FDG PET is an early indicator for metabolic changes. It appears to be a useful add-on to improve accuracy of diagnostic criteria in suspected NS without MRI findings.

Plasma Markers for Therapy Response Monitoring in Patients with Neuroendocrine Tumors Undergoing Peptide Receptor Radionuclide Therapy.

BACKGROUND: Pretherapeutic chromogranin A, alkaline phosphatase (ALP), or De Ritis ratio (aspartate aminotransferase/alanine aminotransferase) are prognostic factors in patients with metastatic neuroendocrine tumors (NET) undergoing peptide receptor radionuclide therapy (PRRT). However, their value for intratherapeutic monitoring remains unclear. We evaluated if changes in plasma markers during PRRT can help identify patients with unfavorable outcomes. METHODS: A monocentric retrospective analysis of 141 patients with NET undergoing PRRT with [177Lu]Lu-DOTATOC was conducted. Changes in laboratory parameters were calculated by dividing the values determined immediately before each cycle of PRRT by the pretherapeutic value. Patients with low vs. high PFS were compared with the Wilcoxon rank-sum test. RESULTS: Progression, relapse, or death after PRRT was observed in 103/141 patients. Patients with low PFS showed a significant relative ALP increase before the third (p = 0.014) and fourth (p = 0.039) cycles of PRRT. Kaplan-Meier analysis revealed a median PFS of 24.3 months (95% CI, 20.7-27.8 months) in patients with decreasing ALP values (Δ > 10%) during treatment, 12.5 months (95% CI, 9.2-15.8 months) in patients with increasing ALP values (Δ > 10%), and 17.7 months (95% CI, 13.6-21.8 months) with stable ALP values (Δ ± 10%). CONCLUSIONS: Based on these exploratory data, a rise in plasma ALP might indicate disease progression and should be interpreted cautiously during therapy.

ChatGPT: Can You Prepare My Patients for [18F]FDG PET/CT and Explain My Reports?

We evaluated whether the artificial intelligence chatbot ChatGPT can adequately answer patient questions related to [18F]FDG PET/CT in common clinical indications before and after scanning. Methods: Thirteen questions regarding [18F]FDG PET/CT were submitted to ChatGPT. ChatGPT was also asked to explain 6 PET/CT reports (lung cancer, Hodgkin lymphoma) and answer 6 follow-up questions (e.g., on tumor stage or recommended treatment). To be rated "useful" or "appropriate," a response had to be adequate by the standards of the nuclear medicine staff. Inconsistency was assessed by regenerating responses. Results: Responses were rated "appropriate" for 92% of 25 tasks and "useful" for 96%. Considerable inconsistencies were found between regenerated responses for 16% of tasks. Responses to 83% of sensitive questions (e.g., staging/treatment options) were rated "empathetic." Conclusion: ChatGPT might adequately substitute for advice given to patients by nuclear medicine staff in the investigated settings. Improving the consistency of ChatGPT would further increase reliability.

Score-based improvement of the indication for surgical treatment of thyroid lesions.

BACKGROUND: The prevalence of focal lesions in the thyroid is high in Germany. In 2018 about 70 000 thyroid surgeries were performed, although the malignancy rate of such findings is low. For this reason it is important to conduct an adequate selection of patients for whom surgery is indicated. AIM: The aim of our work was to validate the preoperative indication for surgery of thyroid lesions based on an independent, self-developed clinical score. PATIENTS AND METHODS: The patient data were evaluated retrospectively over the period 2013 to 2014. A prerequisite for inclusion was that the patients had carried out their complete treatment in domo. The multiparametic score was determined retrospectively and ranges from 3 to 15. The subjective improvement of symptoms (self-disclosure > 6 months postoperatively) and the presence of malignant histology were evaluated as positive outcome parameters. RESULTS: From a collective of 180 patients, 36 patients could be included, in whom all score-relevant parameters had been surveyed. The score distribution was 10 % score 3, 12.5 % score 4, 25 % score 5, 25 % score 6, 12.5 % score 7, 7.5 % score 8, 5 % score 9 and 2.5 % score 10. Using ROC analysis shows an AUC of 0.903, which is a very good differentiation. With a CUT-off score of 7 or higher, 86 % of patients have benefited from surgery. CONCLUSION: Our score with the parameters clinical complaints, sonographically defined size of the thyroid and the cytological result of a fine needle biopsy can lead to an improvement of the indication for surgical treatment of thyroid nodules.