eprintid: 13914
rev_number: 2
eprint_status: archive
userid: 1
dir: disk0/00/01/39/14
datestamp: 2023-11-10 03:28:28
lastmod: 2023-11-10 03:28:28
status_changed: 2023-11-10 01:52:17
type: article
metadata_visibility: show
creators_name: Gajbhiye, G.O.
creators_name: Nandedkar, A.V.
creators_name: Faye, I.
title: Automatic report generation for chest X-Ray images: A multilevel multi-attention approach
ispublished: pub
keywords: Behavioral research; Computer vision; Convolutional neural networks; Decoding; Signal encoding, Chest X-ray image; Course information; Medical professionals; Radiology reports; Sequential dependencies; Short term memory; State-of-the-art methods; Time-consuming tasks, Medical imaging
note: cited By 7; Conference of 4th International Conference on Computer Vision and Image Processing, CVIP 2019 ; Conference Date: 27 September 2019 Through 29 September 2019; Conference Code:238879
abstract: A comprehensive X-ray imaging report greatly assists the medical professional to investigate an indispensable condition and medication. The preparation of an extensive and diversified medical report by analysing the chest X-ray image is a time-consuming task and requires highly experienced professionals. This work targets the fundamental problem of generating a long and multifarious medical report for the chest X-ray image. It introduces a novel Multilevel Multi-Attention based encoder-decoder approach by combining Context Level Visual Attention and Textual Attention to generate a plausible medical report for different views of chest X-ray images. It exploited the proven ability of the Convolutional Neural Network to acquire course information of visual-spatial regions as an encoder. It leverages the strength of the Long Short-Term Memory network to learn long sequential dependencies and the ability of attention to focus on the prominent section as a decoder. The proposed method emphasizes on contextual coherence in intra and inter-sentence dependency within a report to improve the overall medical report generation quality. The effectiveness of the proposed model is evaluated on the publicly available IU chest X-ray dataset consisting of chest images along with multifarious radiology reports. The final performance of the proposed model is reported using the COCO-caption evaluation API. It shows a significant improvement in a medical report generation task compared to state-of-the-art methods. © Springer Nature Singapore Pte Ltd 2020.
date: 2020
publisher: Springer
official_url: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85083724536&doi=10.1007%2f978-981-15-4015-8_15&partnerID=40&md5=e28028f91eaaf6e681e9fa1574c112b0
id_number: 10.1007/978-981-15-4015-8₁₅
full_text_status: none
publication: Communications in Computer and Information Science
volume: 1147 C
pagerange: 174-182
refereed: TRUE
isbn: 9789811540141
issn: 18650929
citation:   Gajbhiye, G.O. and Nandedkar, A.V. and Faye, I.  (2020) Automatic report generation for chest X-Ray images: A multilevel multi-attention approach.  Communications in Computer and Information Science, 1147 C.  pp. 174-182.  ISSN 18650929