Optimizing Acute Ischemic Stroke Outcome Prediction by Integrating Radiomics Features of DSC-PWI and Perfusion Parameter Maps

Huihui Yang ^1,2 Yingwei Guo ³ Jiaxi Lu ^1,2 HASSAN HASEEB ¹ Anbo Cao ^1,2 Yingjian Yang ⁴ Mazen M. Yassin ^1,5 Asim Zaman ^1,5 Xueqiang Zeng ^1,2 Xiaoqiang Miao ^1,6 Ziran Chen ^1,6 Guangtao Huang ^1,2 Taiyu Han ^1,2 Huiling Qiu ^7* Yu Luo ^8* Yan Kang ^1,2,5,6,9*

• ¹ College of Health Science and Environmental Engineering, Shenzhen Technology University, Shenzhen, Guangdong Province, China
• ² Country School of Applied Technology, Shenzhen University, Shenzhen, China
• ³ School of Electrical and Information Engineering, Northeast Petroleum University, Daqing, China
• ⁴ Department of Radiological Research and Development, Shenzhen Lanmage Medical Technology Co., Ltd., Shenzhen, China
• ⁵ School of Biomedical Engineering, Shenzhen University Medical School, Shenzhen University, Shenzhen, China
• ⁶ College of Medicine and Biological Information Engineering, Northeastern University, Shenyang, China
• ⁷ College of Pharmacy, Shenzhen Technology University, Shenzhen, Guangdong Province, China
• ⁸ Department of Radiology, Shanghai Fourth People’s Hospital Affiliated to Tongji University School of Medicine, Shanghai, China
• ⁹ Engineering Research Centre of Medical Imaging and Intelligent Analysis, Ministry of Education, Shenyang, China

Accurate prediction of the prognostic outcomes for patients with ischemic stroke can contribute to personalized treatment decisions and improve life-saving outcomes. This study focuses on the performance of critical moments DSC-PWI in the prognostic prediction of acute ischemic stroke (AIS). It aims to integrate this with perfusion parameters to enhance prediction accuracy. Firstly, The radiomics technique employed to extract DSC-PWI features of critical moments and perfusion parameter features. Following this, a T-test and Lasso algorithm was used to select features associated with the prognosis. Subsequently, machine learning techniques were applied to predict the predictive outcomes of AIS patients. The experimental results showed that DSC-PWI sequences at three critical time points-the first moment after contrast injection, the moment of minimum mean time intensity, and the last moment, collectively referred to as 3PWI, had better prognostic prediction than a single perfusion parameter, achieving an optimal model AUC of 0.863. The performance improved by 23.9%, 19.6%, 6%, and 24% compared with CBV, CBF, MTT, and Tmax parameters. The best prognostic prediction for AIS was obtained by integrating the radiomic features from both 3PWI and perfusion parameters, resulting in the highest AUC of 0.915. Therefore, Integrating the radiomics features of DSC-PWI sequences of three critical scan time points with those from perfusion parameters can further improve the accuracy of prognostic prediction for AIS patients. This approach may provide new insights into the prognostic evaluation of AIS and provide clinicians with valuable support in making treatment decisions.