Weakly supervised multitask learning models to identify symptom onset time of unclear-onset intracerebral hemorrhage.

BACKGROUND: Approximately one-third of spontaneous intracerebral hemorrhage patients did not know the onset time and were excluded from studies about time-dependent treatments for hyperacute spontaneous intracerebral hemorrhage. AIMS: To help clinicians explore the benefit of time-dependent treatments for unclear-onset patients, we presented artificial intelligence models to identify onset time using non-contrast computed tomography (NCCT) based on weakly supervised multitask learning (WS-MTL) structure. METHODS: The patients with reliable symptom onset time (strong label) or repeat CT (weak label) were included and split into training set and test set (internal and external). The WS-MTL structure utilized strong and weak labels simultaneously to improve performance. The models included three binary classification models for classifying whether NCCT acquired within 6, 8 or 12 h for different treatments measured by area under curve, and a regression model for determining the exact onset time measured by mean absolute error. The generalizability of models was also explored in comprehensive analysis. RESULTS: A total of 4004 patients with 10,780 NCCT scans were included. The performance of WS-MTL classification model showed high accuracy, and that of regression model was satisfactory in ≤6 h subgroup. In comprehensive analysis, the WS-MTL showed better performance for larger hematomas and thinner scans. And the performance improved effectively as training amounts increasing and could be improved steadily through retraining. CONCLUSIONS: The WS-MTL models showed good performance and generalizability. Considering the large number of unclear-onset spontaneous intracerebral hemorrhage patients, it may be worth to integrate the WS-MTL model into clinical practice to identify the onset time.

The Patterns of Morphological Change During Intracerebral Hemorrhage Expansion: A Multicenter Retrospective Cohort Study.

Objectives: Hemorrhage expansion (HE) is a common and serious condition in patients with intracerebral hemorrhage (ICH). In contrast to the volume changes, little is known about the morphological changes that occur during HE. We developed a novel method to explore the patterns of morphological change and investigate the clinical significance of this change in ICH patients. Methods: The morphological changes in the hematomas of ICH patients with available paired non-contrast CT data were described in quantitative terms, including the diameters of each hematoma in three dimensions, the longitudinal axis type, the surface regularity (SR) index, the length and direction changes of the diameters, and the distance and direction of movement of the center of the hematoma. The patterns were explored by descriptive analysis and difference analysis in subgroups. We also established a prognostic nomogram model for poor outcomes in ICH patients using both morphological changes and clinical parameters. Results: A total of 1,094 eligible patients from four medical centers met the inclusion criteria. In 266 (24.3%) cases, the hematomas enlarged; the median absolute increase in volume was 14.0 [interquartile range (IQR), 17.9] mL. The initial hematomas tended to have a more irregular shape, reflected by a larger surface regularity index, than the developed hematomas. In subtentorial and deep supratentorial hematomas, the center moved in the direction of gravity. The distance of center movement and the length changes of the diameters were small, with median values of less than 4 mm. The most common longitudinal axis type was anterior-posterior (64.7%), and the axis type did not change between initial and repeat imaging in most patients (95.2%). A prognostic nomogram model including lateral expansion, a parameter of morphological change, showed good performance in predicting poor clinical outcomes in ICH patients. Conclusions: The present study provides a morphological perspective on HE using a novel automatic approach. We identified certain patterns of morphological change in HE, and we believe that some morphological change parameters could help physicians predict the prognosis of ICH patients.