The latest quarterly magazine from The Radiological Society of the Netherlands contains three user stories of AI lung nodule solution Veye Lung Nodules. Radiologists at Northwest Clinics (NWZ) and St. Antonius hospital describe their experience deploying and using Veye Lung Nodules in routine clinical practice. The authors highlight the benefits of workflow integration, service, and support in the detection and measurement of lung nodules. Read below a summary of the articles.
Lung nodule management with Veye Lung Nodules. Experiences at NWZ
By Suzanne Tol, Maarten van de Weijver, Floris Rietema
Radiologists at NWZ have been using Veye Lung Nodules since 2019. This article presents the results of an internal survey among radiologists and pulmonologists. The aim was to evaluate the ease of use and gains of the AI lung nodule management solution.
Most respondents (26 out of 31) are happy with using Veye Lung Nodules. They appreciate that the application seamlessly integrates into their workflow and does not require any manual input. 20 interviewees praise the tool:
“[Veye Lung Nodules is] an ideal support and addition in the detection of lung nodules.”
Radiologists believe that Veye Lung Nodules increases sensitivity and saves time during their nodule assessment, especially when comparing multiple lung nodules with a prior scan. In this case, they indicate an average time gain of approximately 10 minutes per scan.
Some respondents find that the results provide a clear image of the total number of nodules and the potential volume growth (including volume doubling time or VDT). They value the 3D volumetry, which follows the British Thoracic Society (BTS) guidelines. The expectation is that using VDT as an indicator of the nodule’s evolution reduces the number of follow-up scans.
Physicians assume that the sensitivity of the software will continue to improve. Yet, they are aware that AI will continue to play a supporting role for the deciding radiologist.
The article concludes that most of the Veye Lung Nodules users are ‘enthusiastic’. They notice an increase in sensitivity and a decrease in reporting time. These benefits do not cause workflow disruptions.
AI implementation stories. Lessons learned in NWZ hospital
By Mohammad H. Rezazade Mehrizi and Paul Algra
The radiology department at NWZ is an early AI adopter. They are interested in testing solutions and providing feedback for further developments. This article focuses on the implementation of Veye Lung Nodules in clinical practice.
Veye Lung Nodules became available for their team within a few months, following defined legal agreements, e.g. around data protection. A lesson learned from this process is the need for a hospital to reach a level of ‘AI maturity’. This includes the coordination of different internal stakeholders and setting legal frameworks. Also, assigning an internal ‘AI champion’ helps drive digital transformation.
Another observation for radiology teams and hospitals weighing AI solutions:
“It’s important to look beyond the algorithm and consider other aspects such as the track-record of the AI vendor, Q&R standards, IT security standards, local support levels, etc.”
The authors note that Veye Lung Nodules currently analyses more than 60% of the chest CT scans performed at NWZ. Although there is no clinical study of its efficiency yet, they report that:
“Using Aidence can save 2 minutes per-case if it shows there is no nodule in the study; 3 minutes per case if only one nodule is detected; around 5 minutes per case if the radiologist needs to compare the change in the nodule by comparing it with the previous examination (follow-up). However, most radiologists think the help of Aidence is indispensable.”
Interview: experiences with Aidence in clinical practice
With Lucianne Langezaal, Wouter van Es, David Meek
Radiologists at St. Antonius Nieuwegein hospital have been using Veye Lung Nodules since May 2020. Dr. van Es explains their decision:
“As a large thorax center, we wished to innovate. The implementation of Aidence offered the opportunity to be at the forefront of deploying AI capabilities. Aidence was a natural choice.”
Veye Lung Nodules users mostly see its added value in the detection of lung nodules and the second, AI-powered read. They often use Veye Lung Nodules as an ‘add-on’ to check that they have not missed any relevant nodules.
Unlike the reports from NWZ, the three physicians at St. Antonius do not believe that Veye Lung Nodules saves them time. Furthermore, Dr. van Es indicates that there is an occasional delay in receiving the results, possibly caused by the scan migration to a new database at the hospital.
The interviewees add that the AI system is not always fully accurate, as specificity values relate to the selected sensitivity threshold. In addition, the hospital has three locations with different scanners and protocols. Sharing feedback and data with Aidence requires time, but helps further improve and tailor the product for the users.
The hospital plans a systematic evaluation of Veye Lung Nodules next year to see if AI-aided readings allow radiologists to miss fewer lung nodules.
Background: NVvR and MemoRad
The Radiological Society of the Netherlands (NVvR) was founded on April 14th, 1901. The Society aims to ensure accessibility to and quality of radiology and promotes a positive attitude towards the radiology profession.
MemoRad is a quarterly magazine for and by members of NVvR. The 2021-1 edition appeared on March 27th and focused on artificial intelligence. The full publication is available here, in Dutch.
*Note: Veye Lung Nodules was previously known as Veye Chest.