JMIR Publications

ABSTRACT

Background:

Early identification of the etiology of spontaneous intracerebral hemorrhage (ICH) could significantly contribute to planning a suitable treatment strategy. A notable radiomics-based artificial intelligence (AI) model for classifying causes of spontaneous ICH from brain computed tomography (CT) scans has been previously proposed.

Objective:

This study aims to externally validate and assess the utility of the AI model.

Methods:

This study used 69 CT scans from a separate cohort to evaluate the AI model’s performance in classifying non-traumatic ICH into primary, tumorous, and vascular malformations. We also assessed the accuracy, sensitivity, specificity, and positive predictive value of clinicians, radiologists and trainees in identifying the ICH causes before and after adopting the model as assistance. The performances were statistically analyzed by specialty and expertise levels.

Results:

The AI model achieved an overall accuracy of 0.65 in classifying the three causes of ICH. The model’s assistance improved overall diagnostic performance, narrowing the gap between non-radiology and radiology groups, as well as between trainees and experts. The accuracy increases from 0.68 to 0.72, 0.72 to 0.76, 0.69 to 0.74 and 0.72 to 0.75 for non-radiology, radiology, trainees, and specialists, respectively. With the model’s support, radiology professionals demonstrated the highest accuracy, highlighting the model’s potential to enhance diagnostic consistency across different levels.

Conclusions:

When applied to an external dataset, the accuracy of the AI model in categorizing spontaneous ICHs based on radiomics decreased. However, employing the model as an assistant substantially improved the performance of all reader groups, including trainees and radiology and non-radiology specialists.