Dr. Harshita Sharma, PhD (Dr.-Ing.) - University of Oxford - Postdoc

Dr. Harshita Sharma

PhD (Dr.-Ing.)

University of Oxford

Postdoc

Oxford, Oxfordshire | United Kingdom

Additional Specialties: Computer Engineering, Biomedical Imaging, Computer vision, Data Science

ORCID logohttps://orcid.org/0000-0003-4683-2606


Top Author

Dr. Harshita Sharma, PhD (Dr.-Ing.) - University of Oxford - Postdoc

Dr. Harshita Sharma

PhD (Dr.-Ing.)

Introduction

https://www.drharshitasharma.com/

Primary Affiliation: University of Oxford - Oxford, Oxfordshire , United Kingdom

Additional Specialties:

Research Interests:

Education

Apr 2017
Technische Universität Berlin
Ph.D (Dr.-Ing.)
Apr 2013
Jaypee Institute of Information Technology
Lecturer
Jun 2012
Indian Institute of Technology Roorkee
M.Tech
Jul 2010
Indira Gandhi Delhi Technical University for Women
B.Tech

Experience

Jun 2013
DAAD PhD scholarship
DAAD scholar
Jan 2012 - Jan 2013
Jaypee Institute of Information Technology
Lecturer
Electronics and Communication Engineering
Sep 2011
IIT-DAAD Master sandwich scholarship
DAAD scholar
May 2017
University of Oxford
Postdoctoral Research Scientist
Institute of Biomedical Engineering

Publications

22Publications

70Reads

312Profile Views

3PubMed Central Citations

Towards Capturing Sonographic Experience: Cognition-Inspired Ultrasound Video Saliency Prediction

vol 1065. Springer, Cham

In: Zheng Y., Williams B., Chen K. (eds) Medical Image Understanding and Analysis. MIUA 2019. Communications in Computer and Information Science

For visual tasks like ultrasound (US) scanning, experts direct their gaze towards regions of task-relevant information. Therefore, learning to predict the gaze of sonographers on US videos captures the spatio-temporal patterns that are important for US scanning. The spatial distribution of gaze points on video frames can be represented through heat maps termed saliency maps. Here, we propose a temporally bidirectional model for video saliency prediction (BDS-Net), drawing inspiration from modern theories of human cognition. The model consists of a convolutional neural network (CNN) encoder followed by a bidirectional gated-recurrent-unit recurrent convolutional network (GRU-RCN) decoder. The temporal bidirectionality mimics human cognition, which simultaneously reacts to past and predicts future sensory inputs. We train the BDS-Net alongside spatial and temporally one-directional comparative models on the task of predicting saliency in videos of US abdominal circumference plane detection. The BDS-Net outperforms the comparative models on four out of five saliency metrics. We present a qualitative analysis on representative examples to explain the model’s superior performance.

View Article
January 2020
1 Read

Efficient Ultrasound Image Analysis Models with Sonographer Gaze Assisted Distillation.

Med Image Comput Comput Assist Interv 2019 10;22(Pt 4):394-402. Epub 2019 Oct 10.

University of Oxford, Oxford OX3 7DQ, United Kingdom.

View Article

Download full-text PDF

Source
http://dx.doi.org/10.1007/978-3-030-32251-9_43DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6962054PMC
October 2019

Efficient Ultrasound Image Analysis Models with Sonographer Gaze Assisted Distillation

Shen D. et al. (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019. MICCAI 2019. Lecture Notes in Computer Science, vol 11767. Springer, Cham

Lecture Notes in Computer Science

Recent automated medical image analysis methods have attained state-of-the-art performance but have relied on memory and compute-intensive deep learning models. Reducing model size without significant loss in performance metrics is crucial for time and memory-efficient automated image-based decision-making. Traditional deep learning based image analysis only uses expert knowledge in the form of manual annotations. Recently, there has been interest in introducing other forms of expert knowledge into deep learning architecture design. This is the approach considered in the paper where we propose to combine ultrasound video with point-of-gaze tracked for expert sonographers as they scan to train memory-efficient ultrasound image analysis models. Specifically we develop teacher-student knowledge transfer models for the exemplar task of frame classification for the fetal abdomen, head, and femur. The best performing memory-efficient models attain performance within 5% of conventional models that are 1000× larger in size.

View Article
October 2019
1 Read

Captioning Ultrasound Images Automatically

Shen D. et al. (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019. MICCAI 2019. Lecture Notes in Computer Science, vol 11767

Lecture Notes in Computer Science

We describe an automatic natural language processing (NLP)-based image captioning method to describe fetal ultrasound video content by modelling the vocabulary commonly used by sonographers and sonologists. The generated captions are similar to the words spoken by a sonographer when describing the scan experience in terms of visual content and performed scanning actions. Using full-length second-trimester fetal ultrasound videos and text derived from accompanying expert voice-over audio recordings, we train deep learning models consisting of convolutional neural networks and recurrent neural networks in merged configurations to generate captions for ultrasound video frames. We evaluate different model architectures using established general metrics (BLEU, ROUGE-L) and application-specific metrics. Results show that the proposed models can learn joint representations of image and text to generate relevant and descriptive captions for anatomies, such as the spine, the abdomen, the heart, and the head, in clinical fetal ultrasound scans.

View Article
October 2019
1 Read

Spatio-Temporal Partitioning and Description of Full- Length Routine Fetal Anomaly Ultrasound Scans

Biomedical Imaging (ISBI 2019) 2019 IEEE 16th International Symposium on, pp. 987-990, 2019.

2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), Venice, Italy

This paper considers automatic clinical workflow description of full-length routine fetal anomaly ultrasound scans using deep learning approaches for spatio-temporal video analysis. Multiple architectures consisting of 2D and 2D + t CNN, LSTM, and convolutional LSTM are investigated and compared. The contributions of short-term and long-term temporal changes are studied, and a multi-stream framework analysis is found to achieve the best top-l accuracy =0.77 and top-3 accuracy =0.94. Automated partitioning and characterisation on unlabelled full-length video scans show high correlation (ρ=0.95, p=0.0004) with workflow statistics of manually labelled videos, suggesting practicality of proposed methods.

View Article
July 2019
6 Reads

Ultrasound Image Representation Learning by Modeling Sonographer Visual Attention

Chung A., Gee J., Yushkevich P., Bao S. (eds) Information Processing in Medical Imaging. IPMI 2019. Lecture Notes in Computer Science, vol 11492. Springer, Cham

Lecture Notes in Computer Science

Image representations are commonly learned from class labels, which are a simplistic approximation of human image understanding. In this paper we demonstrate that transferable representations of images can be learned without manual annotations by modeling human visual attention. The basis of our analyses is a unique gaze tracking dataset of sonographers performing routine clinical fetal anomaly screenings. Models of sonographer visual attention are learned by training a convolutional neural network (CNN) to predict gaze on ultrasound video frames through visual saliency prediction or gaze-point regression. We evaluate the transferability of the learned representations to the task of ultrasound standard plane detection in two contexts. Firstly, we perform transfer learning by fine-tuning the CNN with a limited number of labeled standard plane images. We find that fine-tuning the saliency predictor is superior to training from random initialization, with an average F1-score improvement of 9.6% overall and 15.3% for the cardiac planes. Secondly, we train a simple softmax regression on the feature activations of each CNN layer in order to evaluate the representations independently of transfer learning hyper-parameters. We find that the attention models derive strong representations, approaching the precision of a fully-supervised baseline model for all but the last layer.

http://dx.doi.org/10.1007/978-3-030-20351-1_46

View Article
May 2019
6 Reads

Multi-task SonoEyenet: detection of fetal standardized planes assisted by generated sonographer attention maps

Frangi A., Schnabel J., Davatzikos C., Alberola-López C., Fichtinger G. (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2018. MICCAI 2018. Lecture Notes in Computer Science, vol 11070. Springer, Cham

Lecture Notes in Computer Science

We present a novel multi-task convolutional neural network called Multi-task SonoEyeNet (M-SEN) that learns to generate clinically relevant visual attention maps using sonographer gaze tracking data on input ultrasound (US) video frames so as to assist standardized abdominal circumference (AC) plane detection. Our architecture consists of a generator and a discriminator, which are trained in an adversarial scheme. The generator learns sonographer attention on a given US video frame to predict the frame label (standardized AC plane/background). The discriminator further fine-tunes the predicted attention map by encouraging it to mimick the ground-truth sonographer attention map. The novel model expands the potential clinical usefulness of a previous model by eliminating the requirement of input gaze tracking data during inference without compromising its plane detection performance (Precision: 96.8, Recall: 96.2, F-1 score: 96.5).

https://ora.ox.ac.uk/objects/uuid:0198daa3-fcc0-4b5f-817a-ffdf6bd5e6c7

View Article
September 2018
18 Reads

SonoEyeNet: Standardized fetal ultrasound plane detection informed by eye tracking

2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018), Washington, DC, 2018, pp. 1475-1478.

2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018)

We present a novel automated approach for detection of standardized abdominal circumference (AC) planes in fetal ultrasound built in a convolutional neural network (CNN) framework, called SonoEyeNet, that utilizes eye movement data of a sonographer in automatic interpretation. Eye movement data was collected from experienced sonographers as they identified an AC plane in fetal ultrasound video clips. A visual heatmap was generated from the eye movements for each video frame. A CNN model was built using ultrasound frames and their corresponding visual heatmaps. Different methods of processing visual heatmaps and their fusion with image feature maps were investigated. We show that with the assistance of human visual fixation information, the precision, recall and F1-score of AC plane detection was increased to 96.5%, 99.0% and 97.8% respectively, compared to 73.6%, 74.1% and 73.8% without using eye fixation information.

View Article
May 2018
18 Reads

Deep convolutional neural networks for automatic classification of gastric carcinoma using whole slide images in digital histopathology

Volume 61, November 2017, Pages 2-13

Computerized Medical Imaging and Graphics

Deep learning using convolutional neural networks is an actively emerging field in histological image analysis. This study explores deep learning methods for computer-aided classification in H&E stained histopathological whole slide images of gastric carcinoma. An introductory convolutional neural network architecture is proposed for two computerized applications, namely, cancer classification based on immunohistochemical response and necrosis detection based on the existence of tumor necrosis in the tissue. Classification performance of the developed deep learning approach is quantitatively compared with traditional image analysis methods in digital histopathology requiring prior computation of handcrafted features, such as statistical measures using gray level co-occurrence matrix, Gabor filter-bank responses, LBP histograms, gray histograms, HSV histograms and RGB histograms, followed by random forest machine learning. Additionally, the widely known AlexNet deep convolutional framework is comparatively analyzed for the corresponding classification problems. The proposed convolutional neural network architecture reports favorable results, with an overall classification accuracy of 0.6990 for cancer classification and 0.8144 for necrosis detection.

https://www.sciencedirect.com/science/article/pii/S0895611117300502

View Article
November 2017
19 Reads

A Content-Based Medical Image Mining System for Knowledge Discovery in Medical Images

Proceedings of National Conference on Information Management in Knowledge Economy

View Article
June 2010
21 Reads

Top co-authors

Pierre Chatelain
Pierre Chatelain

Uppsala University

2
Lior Drukker
Lior Drukker

Shaare Zedek Medical Center

2
Olaf Hellwich
Olaf Hellwich

Technical University Berlin

1
Peter Hufnagl
Peter Hufnagl

Charité-Universitätsmedizin

1
Norman Zerbe
Norman Zerbe

Institute of Pathology

1
Peter Leskovsky
Peter Leskovsky

Computer Vision Lab

1
Aris T Papageorghiou
Aris T Papageorghiou

University of London

1
Yifan Cai
Yifan Cai

Brain Hospital

1
Aris Papageorghiou
Aris Papageorghiou

University of London

1