ABSTRACT
Background Large language models (LLMs) hold substantial promise for medical imaging interpretation. However, there is a lack of studies on their feasibility in handling reasoning questions associated with medical diagnosis. Purpose To investigate the viability of leveraging three publicly available LLMs to enhance consistency and diagnostic accuracy in medical imaging based on standardized reporting, with pathology as the reference standard. Materials and Methods US images of thyroid nodules with pathologic results were retrospectively collected from a tertiary referral hospital between July 2022 and December 2022 and used to evaluate malignancy diagnoses generated by three LLMs-OpenAI's ChatGPT 3.5, ChatGPT 4.0, and Google's Bard. Inter- and intra-LLM agreement of diagnosis were evaluated. Then, diagnostic performance, including accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC), was evaluated and compared for the LLMs and three interactive approaches: human reader combined with LLMs, image-to-text model combined with LLMs, and an end-to-end convolutional neural network model. Results A total of 1161 US images of thyroid nodules (498 benign, 663 malignant) from 725 patients (mean age, 42.2 years ± 14.1 [SD]; 516 women) were evaluated. ChatGPT 4.0 and Bard displayed substantial to almost perfect intra-LLM agreement (κ range, 0.65-0.86 [95% CI: 0.64, 0.86]), while ChatGPT 3.5 showed fair to substantial agreement (κ range, 0.36-0.68 [95% CI: 0.36, 0.68]). ChatGPT 4.0 had an accuracy of 78%-86% (95% CI: 76%, 88%) and sensitivity of 86%-95% (95% CI: 83%, 96%), compared with 74%-86% (95% CI: 71%, 88%) and 74%-91% (95% CI: 71%, 93%), respectively, for Bard. Moreover, with ChatGPT 4.0, the image-to-text-LLM strategy exhibited an AUC (0.83 [95% CI: 0.80, 0.85]) and accuracy (84% [95% CI: 82%, 86%]) comparable to those of the human-LLM interaction strategy with two senior readers and one junior reader and exceeding those of the human-LLM interaction strategy with one junior reader. Conclusion LLMs, particularly integrated with image-to-text approaches, show potential in enhancing diagnostic medical imaging. ChatGPT 4.0 was optimal for consistency and diagnostic accuracy when compared with Bard and ChatGPT 3.5. © RSNA, 2024 Supplemental material is available for this article.
Subject(s)
Thyroid Nodule , Humans , Female , Adult , Thyroid Nodule/diagnostic imaging , Retrospective Studies , Language , Neural Networks, Computer , ROC CurveABSTRACT
Background Noninvasive tests can be used to screen patients with chronic liver disease for advanced liver fibrosis; however, the use of single tests may not be adequate. Purpose To construct sequential clinical algorithms that include a US deep learning (DL) model and compare their ability to predict advanced liver fibrosis with that of other noninvasive tests. Materials and Methods This retrospective study included adult patients with a history of chronic liver disease or unexplained abnormal liver function test results who underwent B-mode US of the liver between January 2014 and September 2022 at three health care facilities. A US-based DL network (FIB-Net) was trained on US images to predict whether the shear-wave elastography (SWE) value was 8.7 kPa or higher, indicative of advanced fibrosis. In the internal and external test sets, a two-step algorithm (Two-step#1) using the Fibrosis-4 Index (FIB-4) followed by FIB-Net and a three-step algorithm (Three-step#1) using FIB-4 followed by FIB-Net and SWE were used to simulate screening scenarios where liver stiffness measurements were not or were available, respectively. Measures of diagnostic accuracy were calculated using liver biopsy as the reference standard and compared between FIB-4, SWE, FIB-Net, and European Association for the Study of the Liver guidelines (ie, FIB-4 followed by SWE), along with sequential algorithms. Results The training, validation, and test data sets included 3067 (median age, 42 years [IQR, 33-53 years]; 2083 male), 1599 (median age, 41 years [IQR, 33-51 years]; 1124 male), and 1228 (median age, 44 years [IQR, 33-55 years]; 741 male) patients, respectively. FIB-Net obtained a noninferior specificity with a margin of 5% (P < .001) compared with SWE (80% vs 82%). The Two-step#1 algorithm showed higher specificity and positive predictive value (PPV) than FIB-4 (specificity, 79% vs 57%; PPV, 44% vs 32%) while reducing unnecessary referrals by 42%. The Three-step#1 algorithm had higher specificity and PPV compared with European Association for the Study of the Liver guidelines (specificity, 94% vs 88%; PPV, 73% vs 64%) while reducing unnecessary referrals by 35%. Conclusion A sequential algorithm combining FIB-4 and a US DL model showed higher diagnostic accuracy and improved referral management for all-cause advanced liver fibrosis compared with FIB-4 or the DL model alone. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Ghosh in this issue.
Subject(s)
Algorithms , Elasticity Imaging Techniques , Liver Cirrhosis , Humans , Male , Liver Cirrhosis/diagnostic imaging , Middle Aged , Female , Retrospective Studies , Elasticity Imaging Techniques/methods , Adult , Deep Learning , Liver/diagnostic imaging , Liver/pathology , Aged , Ultrasonography/methodsABSTRACT
BACKGROUND: Obstructive sleep apnea (OSA) is associated with increased risk of lung cancer mortality. Nevertheless, little is known about the underlying molecular mechanisms. This research aimed to investigate differentially expressed genes (DEGs) and explore their function in Lewis lung carcinoma (LLC)-bearing mice exposed to chronic intermittent hypoxia (CIH) by transcriptome sequencing. METHODS: Lung cancer tissues in LLC-bearing mice exposed to CIH or normoxia were subjected for transcriptome sequencing to examine DEGs. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway analyses were employed to explore the function of DEGs. To evaluate the prognostic value of DEGs, the Kaplan-Meier survival analysis in combination with Cox proportional hazard model were applied based on The Cancer Genome Atlas. RESULTS: A total of 388 genes with 207 up-regulated and 181 down-regulated genes were differentially expressed between the CIH and normoxia control groups. Bioinformatics analysis revealed that the DEGs were related to various signaling pathways such as chemokine signaling pathway, IL-17 signaling pathway, TGF-ß signaling pathway, transcriptional misregulation in cancer, natural killer cell mediated cytotoxicity, PPAR signaling pathway. In addition, the DEGs including APOL1, ETFB, KLK8, PPP1R3G, PRL, SPTA1, PLA2G3, PCP4L1, NINJ2, MIR186, and KLRG1 were proven to be significantly correlated with poorer overall survival in lung adenocarcinoma. CONCLUSIONS: CIH caused a significant change of gene expression profiling in LLC-bearing mice. The DEGs were found to be involved in various physiological and pathological processes and correlated with poorer prognosis in lung cancer.
Subject(s)
Adenocarcinoma of Lung , Carcinoma, Lewis Lung , Lung Neoplasms , Animals , Mice , Lung Neoplasms/genetics , Transcriptome , Neoplastic Processes , Hypoxia/geneticsABSTRACT
Liver fibrosis is a critical factor in the advancement of nonalcoholic fatty liver disease towards cirrhosis. There is limited research exploring the association between obstructive sleep apnea (OSA) and liver fibrosis among community populations. The present study aimed to assess the association between sleep apnea (SA) and liver fibrosis based on the National Health and Nutrition Examination Survey (NHANES). Data were acquired from NHANES survey cycle 2017 to 2020. We assessed liver fibrosis by the median values of liver stiffness measurement (LSM). The diagnosis of SA was based on participants' response to sleep questionnaire. Univariate and multivariate logistic regression were used to validate the association of SA and liver fibrosis. A total of 7615 participants were included in this study. The LSM level of SA group was significantly higher than non-SA group. The proportion of liver fibrosis in SA group was markedly higher than that in non-SA group (14.0% vs 7.3%, Pâ <â .001). Univariate logistic analysis showed that SA was positively associated with liver fibrosis (ORâ =â 2.068, 95%CIâ =â 1.715-2.494, Pâ <â .001). Further multivariate logistic analysis revealed that SA was independently associated with increased risk of liver fibrosis after adjusting for confounding factors (ORâ =â 1.277, 95%CIâ =â 1.003-1.625, Pâ =â .048). The current study demonstrated an independent association between self-reported SA and increased risk of ultrasound-defined liver fibrosis in community-based sample.
Subject(s)
Liver Cirrhosis , Nutrition Surveys , Ultrasonography , Humans , Male , Female , Liver Cirrhosis/epidemiology , Liver Cirrhosis/diagnostic imaging , Liver Cirrhosis/complications , Middle Aged , Adult , Sleep Apnea Syndromes/epidemiology , Risk Factors , Cross-Sectional Studies , Aged , Sleep Apnea, Obstructive/epidemiology , Sleep Apnea, Obstructive/complicationsABSTRACT
PURPOSE: We retrospectively compared the diagnostic performance of contrast-enhanced ultrasonography (CEUS) and contrast-enhanced computer tomography-magnetic resonance imaging (CT/MRI) for recurrent hepatocellular carcinoma (HCC) after curative treatment. MATERIALS AND METHODS: After curative treatment with 421 ultrasound (US) detected lesions, 303 HCC patients underwent both CEUS and CT/MRI. Each lesion was assigned a Liver Imaging Reporting and Data System (LI-RADS) category according to CEUS and CT/MRI LI-RADS. Receiver-operating characteristic (ROC) curves were computed to determine the optimal diagnosis algorithms for CEUS, CT and MRI. The diagnostic accuracy, sensitivity, specificity, and area under the curve (AUC) were compared between CEUS and CT/MRI. RESULTS: Among the 421 lesions, 218 were diagnosed as recurrent HCC, whereas 203 lesions were diagnosed as benign. In recurrent HCC, CEUS detected more arterial hyperenhancement (APHE) and washout than CT and more APHE than MRI. CEUS yielded better diagnostic performance than CT (AUC: 0.981 vs. 0.958) (p = 0.024) comparable diagnostic performance to MRI (AUC: 0.952 vs. 0.933) (p > 0.05) when using their optimal diagnostic criteria. CEUS missed 12 recurrent HCCs, CT missed one, and MRI missed none. The detection rate of recurrent HCC on CEUS (94.8%, 218/230) was lower than that on CT/MRI (99.6%, 259/260) (p = 0.001). Lesions located on the US blind spots and visualization score C would hinder the ability of CEUS to detect recurrent HCC. CONCLUSION: CEUS demonstrated excellent diagnostic performance but an inferior detection rate for recurrent HCC. CEUS and CT/MRI played a complementary role in the detection and characterization of recurrent HCC.