Neoadjuvant chemotherapy assessment is imperative for prognostication and clinical management of locally advanced gastric cancer. We propose an incremental supervised contrastive learning model (iSCLM), an interpretable artificial intelligence framework integrating pretreatment CT scans and H&E-stained biopsy images, for improved decision-making regarding neoadjuvant chemotherapy. We have constructed and tested iSCLM using retrospective data from 2,387 patients across 10 medical centers and evaluated its discriminative ability in a prospective cohort (132 patients; ChiCTR2300068917). iSCLM achieves areas under receiver operating characteristic curves of 0.846-0.876 across different test cohorts. Computed tomography (CT) and pathological attention heatmaps from Shapley additive explanations and global sort pooling illustrate additional benefits for capturing morphological features through supervised contrastive learning. Specifically, pathological top-ranked tiles exhibit decreased distances to tumor-invasive borders and increased inflammatory cell infiltration in responders compared with non-responders. Moreover, CD11c expression is elevated in responders. The developed interpretable model at the molecular pathology level accurately predicts chemotherapy efficacy.

This study sought to determine whether T2 cardiac magnetic resonance (CMR) can stage both hemorrhagic and nonhemorrhagic myocardial infarctions (MIs). CMR-based staging of MI with or without contrast agents relies on the resolution of T2 elevations in the chronic phase, but whether this approach can be used to stage both hemorrhagic and nonhemorrhagic MIs is unclear. Hemorrhagic (n = 15) and nonhemorrhagic (n = 9) MIs were created in dogs. Multiparametric noncontrast mapping (T1, T2, and T2*) and late gadolinium enhancement (LGE) were performed at 1.5- and 3.0-T at 5 days (acute) and 8 weeks (chronic) post-MI. CMR relaxation values and LGE intensities of hemorrhagic, peri-hemorrhagic, nonhemorrhagic, and remote territories were measured. Histopathology was performed to elucidate CMR findings. T2 of nonhemorrhagic MIs was significantly elevated in the acute phase relative to remote territories (1.5-T: 39.8 ± 12.8%; 3.0-T: 27.9 ± 16.5%; p < 0.0001 for both) but resolved to remote values by week 8 (1.5-T: -0.0 ± 3.2%; p = 0.678; 3.0-T: -0.5 ± 5.9%; p = 0.601). In hemorrhagic MI, T2 of hemorrhage core was significantly elevated in the acute phase (1.5-T: 17.7 ± 10.0%; 3.0-T: 8.6 ± 8.2%; p < 0.0001 for both) but decreased below remote values by week 8 (1.5-T: -8.2 ± 3.9%; 3.0-T: -5.6 ± 6.0%; p < 0.0001 for both). In contrast, T2 of the periphery of hemorrhage within the MI zone was significantly elevated in the acute phase relative to remote territories (1.5-T: 35.0 ± 16.1%; 3.0-T: 24.2 ± 10.4%; p < 0.0001 for both) and remained elevated at 8 weeks post-MI (1.5-T: 8.6 ± 5.1%; 3.0-T: 6.0 ± 3.3%; p < 0.0001 for both). The observed elevation of T2 in the peri-hemorrhagic zone of MIs and the absence of T2 elevation in nonhemorrhagic MIs were consistent with ongoing or absence of histological evidence of inflammation, respectively. Hemorrhagic MIs are associated with persisting myocardial inflammation and edema, which can confound staging of hemorrhagic MIs when T2 elevations alone are used to discriminate between acute and chronic MI. Moreover, given the poor prognosis in patients with hemorrhagic MI, CMR evidence for myocardial hemorrhage with persistent edema may evolve as a risk marker in patients after acute MI.

BACKGROUND AND PURPOSE: The residual blood flow artifact is a critical confounder for MR black-blood thrombus imaging of cerebral venous sinus thrombosis. This study aimed to conduct a validation of a new MR black-blood thrombus imaging technique with enhanced blood signal suppression. MATERIALS AND METHODS: Twenty-six participants (13 volunteers and 13 patients) underwent conventional imaging methods followed by 2 randomized black-blood thrombus imaging scans, with a preoptimized delay alternating with nutation for tailored excitation (DANTE) preparation switched on and off, respectively. The signal intensity of residual blood, thrombus, brain parenchyma, normal lumen, and noise on black-blood thrombus images were measured. The thrombus volume, SNR of residual blood, and contrast-to-noise ratio for residual blood versus normal lumen, thrombus versus residual blood, and brain parenchyma versus normal lumen were compared between the 2 black-blood thrombus imaging techniques. Segmental diagnosis of venous sinus thrombosis was evaluated for each black-blood thrombus imaging technique using a combination of conventional imaging techniques as a reference. RESULTS: In the volunteer group, the SNR of residual blood (11.3 ± 2.9 versus 54.0 ± 23.4, P < .001) and residual blood-to-normal lumen contrast-to-noise ratio (7.5 ± 3.4 versus 49.2 ± 23.3, P < .001) were significantly reduced using the DANTE preparation. In the patient group, the SNR of residual blood (16.4 ± 8.0 versus 75.0 ± 35.1, P = .002) and residual blood-to-normal lumen contrast-to-noise ratio (12.4 ± 7.8 versus 68.8 ± 35.4, P = .002) were also significantly lower on DANTE-prepared black-blood thrombus imaging. The new black-blood thrombus imaging technique provided higher thrombus-to-residual blood contrast-to-noise ratio, significantly lower thrombus volume, and substantially improved diagnostic specificity and agreement with conventional imaging methods. CONCLUSIONS: DANTE-prepared black-blood thrombus imaging is a reliable MR imaging technique for diagnosing cerebral venous sinus thrombosis.

This study reviewed the serial real‐time reverse‐transcription polymerase chain reaction (rRT‐PCR) results of 37 patients admitted to our hospital in Wuhan, China, who had three or more sequential negative results before discharge. Of these 37 patients, 14 (~38%) had a positive rRT‐PCR result after a negative result during convalescence, and 5 (~14%) had a positive rRT‐PCR result after two consecutive negative results during convalescence. These results suggest that it may be necessary to require that patients have three consecutive negative results before discharge, to ensure that they do not spread infection among members of their household, or in the community. We believe that our study makes a significant contribution to the literature because it is not currently the standard of care to require patients to have three consecutive negative results before discharge. Our results suggest that a relatively high proportion of patients may continue to shed severe acute respiratory syndrome coronavirus 2 after they have clinically recovered, and thus may transmit the infection to others.