Fetal Brain MRI Measurements Using a Deep Learning Landmark Network with Reliability Estimation

Netanell Avisdris, Dafna Ben-Bashat, Liat Ben-Sira, Leo Joskowicz

Abstract: We present a new deep learning method, FML, that automatically computes linear measurements in a fetal brain MRI volume. The method is based on landmark detection and estimates their location reliability. It consists of four steps: 1) fetal brain region of interest detection with a two-stage anisotropic U-Net; 2) reference slice selection with a convolutional neural network (CNN); 3) linear measurement computation based on landmarks detection using a novel CNN, FMLNet; 4) measurement reliability estimation using a Gaussian Mixture Model. The advantages of our method are that it does not rely on heuristics to identify the landmarks, that it does not require fetal brain structures segmentation, and that it is robust since it incorporates reliability estimation. We demonstrate our method on three key fetal biometric measurements from fetal brain MRI volumes: Cerebral Biparietal Diameter (CBD), Bone Biparietal Diameter (BBD), and Trans Cerebellum Diameter (TCD). Experimental results on training $$(N=164)$$ and test $$(N=46)$$ datasets of fetal MRI volumes yield a 95% confidence interval agreement of 3.70 mm, 2.20 mm and 2.40 mm for CBD, BBD and TCD, in comparison to measurements performed by an expert fetal radiologist. All results were below the interobserver variability, and surpass previously published results. Our method is generic, as it can be directly applied to other linear measurements in volumetric scans and can be used in a clinical setup.