The diagnostic performance of ss-EPI-DWI, rs-EPI-DWI (RESOLVE) and TGSE-BLADE-DWI in a model of retinal ischemia: a comparative phantom study

This study systematically compared the performance of ss-EPI-DWI, rs-EPI-DWI (RESOLVE), and TGSE-BLADE-DWI for detecting small ischemic lesions (10 -2 mm) using a DWI phantom to gain insights for optimization of microstructural orbital MRI for retinal ischemia diagnosis. A DWI phantom simulating ischemic lesions embedded in healthy neuronal tissue was used to quantitatively and qualitatively assess the sequences at 3T. Signal-to-noise-ratio (SNR), contrast-to-noise-ratio (CNR), relative contrast (ReCon), geometric distortion rate (GDR) and apparent diffusion coefficient (ADC) accuracy were measured, with correlation analysis examining the effect of lesion size. Qualitative analysis was performed by two neuroradiologists evaluating the images in different criteria using a 5-point Likert scale. Among all sequences rs-EPI-DWI showed the most precise ADC (p < 0.001), highest CNR (p < 0.001) and relative contrast (p < 0.001) and non-inferiority to ss-EPI-DWI in SNR (p = 0.06). Both TGSE-BLADE-DWI and rs-EPI-DWI showed less GDR and susceptibility artifacts than ss-EPI-DWI qualitatively. All sequences depicted even the 2 mm lesion qualitatively; however, lesion size most strongly affected TGSE-BLADE-DWI quantitatively. Therefore, rs-EPI-DWI and TGSE-BLADE-DWI are promising alternatives to conventional ss-EPI-DWI for microstructural orbital MRI and may prove useful for clinical diagnosis of acute retinal ischemia due to high spatial resolution with reliable ADC acquisition, low geometric distortion and reduced susceptibility artifacts.