Chandni Gupta

TL;DR: This work proposes a new Patch-based Iterative Network (PIN), a multitask learning framework that combines regression and classification to improve localisation accuracy in 3D medical volumes and extends PIN to localise multiple landmarks by using principal component analysis, which models the global anatomical relationships between landmarks.

TL;DR: This work proposes a new Iterative Transformation Network (ITN) for the automatic detection of standard planes in 3D volumes and introduces additional classification probability outputs to the network to act as confidence measures for the regressed transformation parameters in order to further improve the localisation accuracy.

TL;DR: In this article, an Iterative Transformation Network (ITN) was proposed to detect standard scan planes in 3D volumes of fetal brain ultrasound. But the standard plane detection in 3-D volume is a labour-intensive task and requires expert knowledge of fetal anatomy.

TL;DR: A two-stage convolutional neural network able to incorporate additional contextual and structural information into the segmentation process in fetal 3DUS, significantly outperforming traditional 2D biometrics.

TL;DR: In this article, a patch-based iterative network (PIN) is proposed for fast and accurate landmark localisation in 3D medical volumes, where patches are repeatedly passed to the CNN until the estimated landmark position converges to the true landmark location.