Project DAME

Use-Lab GmbH has entered into a partnership with COSMONiO IMAGING B.V., UMC Groningen (UMCG), Radiologie Westmünsterland and Carl von Ossietzky University Oldenburg. Within this project called "DAME" (Deep-Learning Algorithm for Medical Imaging Evaluation) Use-lab is responsible for the usability engineering of a deep learning algorithm for the evaluation of medical images.The first "mission" of Use-Lab was the user requirements analysis: literature research, field study and semi-structured interview.

The input for the literature search was a study overview, which was made available to all project partners by Use-Lab to show which actions Use-Lab had planned to carry out for which purpose within the user requirements analysis.

Literature study during the user requirement analysis

A literature review is a search and evaluation of the available literature in a given subject or chosen topic area. A literature search is a search and evaluation of available literature on a specific topic or selected subject area. For the present study, the literature search was conducted in advance in order to provide researchers with a more comprehensive knowledge of medical imaging, its basic principles and any differences between Germany and the Netherlands.

This method was used in May 2018. Research was conducted in scientific journals, the Code of Law, and the Internet. The sources were national statistics, laws and guidelines, scientific journals, manufacturer websites and, if available, reports and documentation of existing medical imaging evaluation systems.

The input for the literature study was a study outline provided by Use-Lab to all project partners, which called for an examination of the current situation regarding computer-based systems for medical imaging evaluation.

After conducting the studies, the gathered data were analyzed and enriched with data from Use-Lab's clinical expertise and general user-centered design information and guidelines. The data analysis phase led to answers to certain open topics and questions of the input documents. All other topics which might not be answered or addressed during this first phase of the usability engineering process will be considered during the further steps in the development process.

Contextual inquiries and semi-structured interviews during the user requirements analysis

A contextual inquiry is the obvious choice for gathering field data from users in the course of their normal activities at their actual workplaces. It is an interaction in which the researcher watches the user's activities as unobtrusively as possible and discusses those activities later with the user. This allows the researcher to get a genuine idea of the setting and environment where the product potentially will be used in the future.

A semi-structured interview is a research method used to gather focused, qualitative data. This method can uncover rich descriptive data from the perspective of end-users or potential end-users. It is an open interview, offering users the possibility to self-report on the use environment and help researchers understand their circumstances. While a structured interview has a rigorous set of questions which does not allow one to divert from, the semi-structured interview is open and has a general framework of themes to be explored. It allows to divert from the original set of questions and for new ideas to be brought up during the interview as a result of what the interviewee says.

For the present study this provided the necessary degree of freedom to really gather comprehensive insights into the medical imaging workflow. Semi-structured interviews are the most used type of interview in qualitative research, where one does not aim for scientific accuracy but for true exploration of the interviewee's opinion, in this case, needs and requrements.

Summary of results

Summarised, three sites in total has been visited in the Netherlands and Germany.

Among other things, information was gathered on the following topics: workflows, image generation, image acquisition, findings and reports, workflow in and after radiology, software experience of the relevant experts and DICOM data structures. Different approaches as well as various detailed steps such as mapping and contouring were also discussed and included in the results.

Videoanimation of the workflow in the radiology

In addition to the user requirements analysis, Use-Lab, in cooperation with a subcontractor, created an animated film that explains the function of DAME in a simple and understandable way. You can find the result here:

Formative Study

During 2020 Use-Lab planned an formative study to improve usability and safety of a graphical user interface (GUI) of a deep-learning algorithm for medical imaging evaluation from potential users (radiologists, medical physicists and MRTA). The algorithm was developed by COSMONiO BV from Groningen. The product is called NOUS, from the Greek word for "mind".

NOUS is an AI platform that can be taught to detect certain patterns in 2D medical imaging data. It is intended to be used by radiologists, who mark anomalies in medical imaging data, as well as medical physicists, who plan radiation therapy by segmenting healthy and unhealthy tissue in medical imaging data.

The formative study was conducted as an expert review by Use-Lab's interdisciplinary human factors experts' team which consists of application specialists, engineers, usability experts and designers. The GUI was presented to the experts. In a collective brainstorming, the GUI was discussed.

Additionally, the expert review was extended by interviewing one radiologist and two medical physicists.

The goal of the expert review was to show strengths and weaknesses of the GUI, mainly focusing on the usability of the GUI for training the algorithm to detect certain tissues in CT-scans. Additionally, aspects like look & feel of the product and the general efficiency of use was considered.

Based on the expertise and feedback of external experts, recommendations for the design of the GUI were developed. Based on these recommendations, a completely revised GUI was developed, which aimed at integrating the recommendations and merging them with each other, as well as orienting more strongly towards the usability heuristics according to Nielsen, where appropriate.

 

This project is financially supported within the INTERREG program by the European Union and the INTERREG partners. WWW.DEUTSCHLAND-NEDERLAND.EU

Funding Partners

 

Project Partners