Project Details

Clinical Integration of Fetal Brain MRI Slice-to-Volume Reconstruction to Improve Anomaly Detection

Arman Avesta, MD, PhD

Summary

BACKGROUND: Fetal brain images are often low-quality and in non-standard planes due to fetal movement during imaging, which can result in missed diagnoses of fetal brain disorders. GAP: To improve the detection of fetal brain anomalies, there is a need to rapidly reconstruct high-quality coherent 3D images of the fetal brain after MR image acquisition. To have clinical impact, these reconstructed fetal brain images should then be returned to the clinical Picture Archiving and Communication System (PACS) for interpretation and clinical decision-making. HYPOTHESIS: Aim 1: Integrate the image quality-assessment and super-resolution slice-to-volume reconstruction (SVR) algorithms into the clinical radiology workflow. This integration will make high-quality coherent 3D fetal brain images available in PACS within minutes of image acquisition. Aim 2: Evaluate whether the availability of high-quality, 3D fetal brain images in PACS improves the detection of fetal brain anomalies. Hypothesis: The availability of these images will increase the sensitivity, specificity, and speed of detecting fetal brain anomalies. METHODS: I will integrate the ResNet quality assessment and SVR algorithms into PACS to generate high-resolution 3D fetal brain images and make them accessible for interpretation. I will then evaluate whether the sensitivity, specificity, and speed of detecting fetal brain disorders improve with the interpretation of these high-quality 3D images. RESULTS: Pending. IMPACT: The generation of high-quality 3D images of the fetal brain and their integration into the radiology workflow can improve the detection of fetal brain anomalies. Website Link: https://connects.catalyst.harvard.edu/Profiles/display/Person/214522