Improving interobserver agreement and performance of deep learning models for segmenting acute ischemic stroke by combining DWI with optimized ADC thresholds

Objectives To examine the role of ADC threshold on agreement across observers and deep learning models (DLMs) plus segmentation performance of DLMs for acute ischemic stroke (AIS). Methods Twelve DLMs, which were trained on DWI-ADC-ADC combination from 76 patients with AIS using 6 different ADC thresholds with ground truth manually contoured by 2 observers, were tested by additional 67 patients in the same hospital and another 78 patients in another hospital. Agreement between observers and DLMs were evaluated by Bland-Altman plot and intraclass correlation coefficient (ICC). The similarity between ground truth (GT) defined by observers and between automatic segmentation performed by DLMs was evaluated by Dice similarity coefficient (DSC). Group comparison was performed using the Mann-Whitney U test. The relationship between the DSC and ADC threshold as well as AIS lesion size was evaluated by linear regression analysis. A p < .05 was considered statistically significant. Results Excellent interobserver agreement and intraobserver repeatability in the manual segmentation (all ICC > 0.98, p < .001) were achieved. The 95% limit of agreement was reduced from 11.23 cm(2) for GT on DWI to 0.59 cm(2) for prediction at an ADC threshold of 0.6 x 10(-3) mm(2)/s combined with DWI. The segmentation performance of DLMs was improved with an overall DSC from 0.738 +/- 0.214 on DWI to 0.971 +/- 0.021 on an ADC threshold of 0.6 x 10(-3) mm(2)/s combined with DWI. Conclusions Combining an ADC threshold of 0.6 x 10(-3) mm(2)/s with DWI reduces interobserver and inter-DLM difference and achieves best segmentation performance of AIS lesions using DLMs.

Automated summary

The objective of this study was to investigate the role of ADC threshold on agreement across observers and deep learning models (DLMs) as well as segmentation performance for acute ischemic stroke (AIS). The results showed that combining ADC threshold with DWI can reduce the differences between observers and DLMs and achieve the best segmentation performance for AIS lesions.