Detection and Grading of Radiographic Hand Osteoarthritis Using an Automated Machine Learning Platform.

OBJECTIVE: Automated machine learning (autoML) platforms allow health care professionals to play an active role in the development of machine learning (ML) algorithms according to scientific or clinical needs. The aim of this study was to develop and evaluate such a model for automated detection and grading of distal hand osteoarthritis (OA). METHODS: A total of 13,690 hand radiographs from 2,863 patients within the Swiss Cohort of Quality Management (SCQM) and an external control data set of 346 non-SCQM patients were collected and scored for distal interphalangeal OA (DIP-OA) using the modified Kellgren/Lawrence (K/L) score. Giotto (Learn to Forecast [L2F]) was used as an autoML platform for training two convolutional neural networks for DIP joint extraction and subsequent classification according to the K/L scores. A total of 48,892 DIP joints were extracted and then used to train the classification model. Heatmaps were generated independently of the platform. User experience of a web application as a provisional user interface was investigated by rheumatologists and radiologists. RESULTS: The sensitivity and specificity of this model for detecting DIP-OA were 79% and 86%, respectively. The accuracy for grading the correct K/L score was 75%, with a κ score of 0.76. The accuracy per DIP-OA class differed, with 86% for no OA (defined as K/L scores 0 and 1), 71% for a K/L score of 2, 46% for a K/L score of 3, and 67% for a K/L score of 4. Similar values were obtained in an independent external test set. Qualitative and quantitative user experience testing of the web application revealed a moderate to high demand for automated DIP-OA scoring among rheumatologists. Conversely, radiologists expressed a low demand, except for the use of heatmaps. CONCLUSION: AutoML platforms are an opportunity to develop clinical end-to-end ML algorithms. Here, automated radiographic DIP-OA detection is both feasible and usable, whereas grading among individual K/L scores (eg, for clinical trials) remains challenging.

The combination of non-contrast abbreviated MRI and alpha foetoprotein has high performance for hepatocellular carcinoma screening.

OBJECTIVES: This study aimed to compare two abbreviated MRI (AMRI) protocols to complete MRI for HCC detection: non-contrast (NC)-AMRI without/with alpha foetoprotein (AFP) and dynamic contrast-enhanced (Dyn)-AMRI. METHODS: This retrospective single-center study included 351 patients (M/F: 264/87, mean age: 57y) with chronic liver disease, who underwent MRI for HCC surveillance between 2014 and 2020. Two reconstructed AMRI sets were obtained based on complete MRI: NC-AMRI (T2-weighted imaging (WI) + diffusion-WI) and Dyn-AMRI (T2-WI + dynamic T1-WI) and were assessed by 2 radiologists who reported all suspicious lesions, using LI-RADS/adapted LI-RADS classification. The reference standard was based on all available patient data. Inter-reader agreement was assessed and MRI diagnostic performance was compared to the reference standard. RESULTS: The reference standard demonstrated 83/351 HCC-positive patients (prevalence: 23.6%, median size: 22 mm, and positive MRIs: 83/631). Inter-reader agreement was substantial for all sets. Sensitivities of Dyn-AMRI and complete MRI (both 92.8%) were similar, higher than NC-AMRI (72.3%, p < 0.001). Specificities were not different between sets. NC-AMRI + AFP (92.8%) had similar sensitivity to Dyn-AMRI and complete MRI. In patients with small size HCCs (≤ 2 cm), sensitivities of Dyn-AMRI (85.3%) and complete MRI (88.2%) remained similar (p = 0.564), also outperforming NC-AMRI (52.9%, p < 0.05). NC-AMRI + AFP had similar sensitivity (88.2%) to Dyn-AMRI and complete MRI (p = 0.706 and p = 1, respectively). CONCLUSIONS: Dyn-AMRI has similar diagnostic performance to complete MRI for HCC detection, while both outperform NC-AMRI, especially for small size HCCs. NC-AMRI + AFP demonstrates similar sensitivity to Dyn-AMRI and complete MRI. CLINICAL RELEVANCE STATEMENT: Due to the low sensitivity of ultrasound for hepatocellular screening, new screening methods are needed. Abbreviated MRI (AMRI) is a candidate, especially non-contrast AMRI with serum alpha foetoprotein as the acquisition time is low, without the need for contrast medium injection. KEY POINTS: • Dynamic contrast-enhanced abbreviated MRI using extracellular gadolinium-based contrast agent and complete MRI have similar diagnostic performance for hepatocellular carcinoma detection in an at-risk population. • Non-contrast abbreviated MRI with alpha foetoprotein has similar diagnostic performance to dynamic contrast-enhanced abbreviated MRI and complete MRI, including when considering small size hepatocellular carcinoma ≤ 2 cm. • Non-contrast abbreviated MRI and dynamic contrast-enhanced abbreviated MRI can be performed in 7 and 10 min, excluding patient setup time.

Intradialytic hypotension and relationship with cognitive function and brain morphometry.

BACKGROUND: The haemodynamic stress brought about by dialysis could justify the loss of structural and functional integrity of the central nervous system (CNS). The main objective of this study was to analyse the relationship between intradialytic hypotension (IDH) and cognitive function and brain morphometry. METHODS: The cross-sectional KIDBRAIN study (Cohort Study of Morphological Changes of the Brain by MRI in Chronic Kidney Disease Patients) included 68 prevalent patients with no history of neurological disorders (cerebrovascular disease and cognitive impairment) undergoing haemodialysis (HD). We analysed 18 non-consecutive dialysis sessions (first three of each month over a 6-month period) and various definitions of IDH were recorded. Global cognitive function (GCF) was assessed using the Mini-Mental State Examination (MMSE) and parameters of structural integrity of the CNS were obtained using volume morphometry magnetic resonance imaging analysis [grey matter (GM), white matter (WM) and hippocampus). RESULTS: A greater number of sessions with IDH were associated with less volume of WM (r = -0.359,P = 0.003) and hippocampus (r = -0.395, P = 0.001) independent of cardiovascular risk factors according to multivariable linear regression models (ß = -0.198, P = 0.046 for WM; ß = -0.253, P = 0.017 for hippocampus). The GCF by the MMSE was 27.3 ± 7.3.1 and was associated with WM volume (ß = 0.403, P = 0.001) independent of GM and hippocampus volume. Symptomatic IDH was associated with GCF (r = -0.420, P < 0.001) in adjusted analysis (ß = -0.339, P = 0.008). CONCLUSIONS: Even when asymptomatic, IDH is associated with a lower WM and hippocampus volume and reduced GCF in patients undergoing HD, thus suggesting greater vulnerability of the brain to the haemodynamic stress that may be generated by a dialysis session.

Gyral and sulcal cortical thinning in adolescents with first episode early-onset psychosis.

BACKGROUND: Psychosis is associated with volumetric decreases of cortical structures. Whether these volumetric decreases imply abnormalities in cortical thickness, surface, or cortical folding is not clear. Due to differences in cytoarchitecture, cortical gyri and sulci might be differentially affected by psychosis. Therefore, we examined differences in gyral and sulcal cortical thickness, surface, folding, and volume between a minimally treated male adolescent population with early-onset first-episode psychosis (EOP) and a healthy control group, with surface-based morphometry. METHODS: Magnetic resonance imaging brain scans were obtained from 49 adolescent EOP patients and 34 healthy control subjects. Subjects were younger than 18 years (age range 12 years-18 years), and EOP patients had a duration of positive symptoms of <6 months. RESULTS: Early-onset first-episode psychosis was associated with local bilateral cortical thinning and volume deficits in both the gyri and sulci of the superior temporal cortex and the inferior, middle, medial, and superior prefrontal cortex. In the pars triangularis and opercularis cortex of patients, gyral cortical thickness was thinner, whereas sulcal thickness was not. Patients exhibited cortical thinning together with a decreased degree of cortical folding in the right superior frontal cortex. CONCLUSIONS: Cortical thinning of both gyri and sulci seem to underlie most cortical volume deficits in adolescent patients with EOP. Except for the right superior frontal region, the degree of cortical folding was normal in regions showing decreased cortical thickness, suggesting that the process of cortical thinning in adolescent patients with EOP primarily takes place after the formation of cortical folds.