Smart Ultrasonography Assistant
We are looking forward to having the split processing computational power of OMAP4 for developing an add-on module which plugs on to a conventional ultrasonography equipment, analyzing video streaming over the video-out ports/network and providing Radiologists/Sonologists with diagnostic assistance based on computer vision and decision support, displayed over an adjunct interactive video display unit (touch screen LCD panel). The final product will have Image Processing apps working on DSP, base OS functioning on ARM and graphics rendering carried out on SGX.
Ultrasonography is an important noninvasive clinical imaging modality preferred for screening and diagnosis of internal organ abnormalities. At the Medical Imaging and PACS Lab, School of Medical Science and Technology, IIT Kharagpur, we are working on an ambition project towards development of "Computer Vision based Approach for Interpretation of Sonomammography".
This work addresses development of a computer vision based approach which seeks to overcome limitations in respect of inter-observer and intra-observer variability of lesion description, feature analysis and final assessment; and resolving overlap of sonomammography features characterizing different grades of tissue abnormality manifested through architectural appearance and sonological nature of the tissue. The developed methods and algorithms being incorporated into an expert system coupled with an ultrasonic equipment will aid in speedy and accurate assessment of breast tissue abnormality, by radiologists’, through quantitative analysis of the images. You can find regular updates about the development at http://debdoot.web.officelive.com/phdthesis.aspx
Currently, we are working with an OMAP-L138 based platform for implementing part of the work on front-end development. Details are available at http://hawkboard.wikispot.org/Front_Page
However our preliminary housekeeping of total computational requirements shows OMAP4 to be optimally satisfactory, and PandaBoard natively provides the minimal hardware resources required to undertake the complete implementation as an embedded solution.