US-based Sequential Algorithm Integrating an AI Model for Advanced Liver Fibrosis Screening.

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.

Collaborative Enhancement of Consistency and Accuracy in US Diagnosis of Thyroid Nodules Using Large Language Models.

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.

Value of Multimodal Data From Clinical and Sonographic Parameters in Predicting Recurrence of Hepatocellular Carcinoma After Curative Treatment.

OBJECTIVE: The objective of the work described here was to assess the value of the combination of pre-operative multimodal data-including clinical data, contrast-enhanced ultrasound (CEUS) information and liver stiffness measurement (LSM) based on 2-D shear wave elastography (SWE)-in predicting early (within 1 y) and late (after 1 y) recurrence of hepatocellular carcinoma (HCC) after curative treatment. METHODS: We retrospectively included 101 patients with HCC who met the Milan criteria and received curative treatment. The multimodel data from clinical parameters, LSM by 2-D SWE and CEUS enhancement patterns were collected. The association between different variables in HCC recurrence was accessed using a Cox proportional hazard model. On the basis of the independent factors of early recurrence, models with different source variables were established (Clinical Model, CEUS-Clinical Model, SWE-Clinical Model, CEUS-SWE-Clinical Model). The goodness-of-fit of models was evaluated and the performance trends of different models were calculated by time-dependent area under the curve (AUC). RESULTS: Two-dimensional SWE, CEUS enhancement patterns and clinical parameters (spleen length, multiple tumors, α-fetoprotein, albumin and prothrombin time) were independently associated with early recurrence (all p values <0.05). Multiple tumors and a decrease in albumin independently contributed to the late recurrence. The model fit of CEUS-SWE-Clinical Model was superior to other models in predicting early recurrence (all p values <0.05). The AUCs of the CEUS-Clinical Model were higher from 2 mo to 7 mo, while the SWE-Clinical Model had higher AUCs from 9 mo to 12 mo. CONCLUSION: CEUS enhancement patterns and 2-D SWE were independent predictors of HCC early recurrence as the two factors contributed to the predictive performance at different times. The multimodal model, which included diverse data in predicting early HCC recurrence, had the best goodness-of-fit.

Performance and comparison of artificial intelligence and human experts in the detection and classification of colonic polyps.

OBJECTIVE: The main aim of this study was to analyze the performance of different artificial intelligence (AI) models in endoscopic colonic polyp detection and classification and compare them with doctors with different experience. METHODS: We searched the studies on Colonoscopy, Colonic Polyps, Artificial Intelligence, Machine Learning, and Deep Learning published before May 2020 in PubMed, EMBASE, Cochrane, and the citation index of the conference proceedings. The quality of studies was assessed using the QUADAS-2 table of diagnostic test quality evaluation criteria. The random-effects model was calculated using Meta-DISC 1.4 and RevMan 5.3. RESULTS: A total of 16 studies were included for meta-analysis. Only one study (1/16) presented externally validated results. The area under the curve (AUC) of AI group, expert group and non-expert group for detection and classification of colonic polyps were 0.940, 0.918, and 0.871, respectively. AI group had slightly lower pooled specificity than the expert group (79% vs. 86%, P < 0.05), but the pooled sensitivity was higher than the expert group (88% vs. 80%, P < 0.05). While the non-experts had less pooled specificity in polyp recognition than the experts (81% vs. 86%, P < 0.05), and higher pooled sensitivity than the experts (85% vs. 80%, P < 0.05). CONCLUSION: The performance of AI in polyp detection and classification is similar to that of human experts, with high sensitivity and moderate specificity. Different tasks may have an impact on the performance of deep learning models and human experts, especially in terms of sensitivity and specificity.

[Neuropsychological development of human immunodeficiency virus-exposed uninfected infants/young children]. / äººç±»å ç–«ç¼ºé™·ç— æ¯’æš´éœ²æœªæ„ŸæŸ“å©´å¹¼å„¿çš„ç¥žç»å¿ƒç†å‘è‚²çŠ¶å†µ.

OBJECTIVES: To investigate the level of neuropsychological development in human immunodeficiency virus (HIV)-exposed uninfected (HEU) infants/young children and the influence of maternal HIV infection on the neuropsychological development of HEU infants/young children. METHODS: A total of 141 HEU infants/young children, aged 0-18 months and born to HIV-infected mothers, who were managed in four maternal and child health care hospitals in Yunnan Province of China from June 2019 to December 2020 and met the inclusion criteria were enrolled as the HEU group. A total of 141 HIV-unexposed uninfected (HUU) infants/young children who were born to healthy mothers and managed in the same hospitals, matched at a ratio of 1:1 based on sex, age, method of birth, birth weight, and gestational age, were enrolled as controls. Griffiths Development Scales-Chinese Edition was used to assess the development in the five domains of locomotion, personal-social, hearing and language, eye-hand co-ordination, and performance (visual perception and space integration ability). A questionnaire survey was performed to collect relevant information. The multivariate logistic regression analysis was used to assess the influence of maternal HIV infection on the neuropsychological development of HEU infants/young children. RESULTS: Compared with the HUU group, the HEU group had significantly higher detection rates of retardation in the domains of hearing and language and performance (P<0.05). The multivariate logistic regression analysis showed that maternal HIV infection increased the risk of retardation in the domains of hearing and language (OR=2.661, 95%CI: 1.171-6.047, P<0.05) and performance (OR=2.321, 95%CI: 1.156-4.658, P<0.05). CONCLUSIONS: Maternal HIV infection can negatively affect the development of hearing and language and performance in HEU infants/young children, and further studies are needed to clarify related mechanisms.

Can monodisperse microbubble-based three-dimensional contrast-enhanced ultrasound reduce quantitative heterogeneity? An in vitro study.

BACKGROUND: Heterogeneity within the tumor may cause large heterogeneity in quantitative perfusion parameters. Three-dimensional contrast-enhanced ultrasound (3D-CEUS) can show the spatial relationship of vascular structure after post-acquisition reconstruction and monodisperse bubbles can resonate the ultrasound pulse, resulting in the increase in sensitivity of CEUS imaging. OBJECTIVES: To evaluate whether the combination of 3D-CEUS and monodisperse microbubbles could reduce the heterogeneity of quantitative CEUS. MATERIAL AND METHODS: Three in vitro perfusion models with perfusion volume ratio of 1:2:4 were set up. Both quantitative 2D-CEUS and 3D-CEUS were used to acquire peak intensity (PI) with 2 kinds of ultrasound agents. One was a new kind of monodisperse bubbles produced in this study, named Octafluoropropane-loaded cerasomal microbubbles (OC-MBs), the other was SonoVue®. The coefficient of variation (CV) was calculated to evaluate the cross-sectional variability. Pearson's correlation analysis was used to assess the correlation between weighted PIs (average of PIs of 3 different planes) and perfusion ratios. RESULTS: The average CVs of quantitative 3D-CEUS was slightly lower than that of 2D-CEUS (0.41 ±0.17 compared to 0.55 ±0.26, p = 0.3592). As for quantitative 3D-CEUS, the PI of the OC-MBs has shown better stability than that of SonoVue®, but without a significant difference (average CVs: 0.32 ±0.19 compared to 0.50 ±0.10, p = 0.0711). In the 2D-CEUS condition, the average CVs of OC-MBs group and SonoVue® group were 0.68 ±0.15 and 0.41 ±0.17 (p = 0.2747). As for 3D-CEUS condition, using OC-MBs group and SonoVue®, the r-values of the weighted PI and perfusion ratio were 0.8685 and 0.5643, respectively, while that of 2D-CEUS condition were 0.7760 and 0.3513, respectively. CONCLUSIONS: Our in vitro experiments showed that OC-MBs have the potential in acquiring more stable quantitative CEUS value, as compared to the SonoVue® in 3D-CEUS condition. The combination of 3D-CEUS and OC-MBs can reflect perfusion volume more precisely and may be a potential way to reduce quantitative heterogeneity.