PhD defence B. (Beibei) Jiang

Artificial Intelligence for Nodule Management in Lung Cancer Screening: Validation in the UKLS Trial

On Tuesday 8 December 2026 B. Jiang will defend the doctoral thesis titled: Artificial Intelligence for Nodule Management in Lung Cancer Screening: Validation in the UKLS Trial

Promotor
Prof.dr. H.J. de Koning
Promotor
Prof.dr. M. Oudkerk
Co-promotor
Dr. M.A. Heuvelmans
Co-promotor
Dr. C.M. van der Aalst
Date
Tuesday 8 Dec 2026, 15:30 - 17:00
Type
PhD defence
Space
Professor Andries Querido Hall
Building
Education Centre
Location
Erasmus MC
Add to calendar

Below is a brief summary of the dissertation:

Low-dose CT screening can reduce lung cancer mortality through early detection, but reviewing large numbers of screening scans creates a substantial workload for radiologists. Most detected pulmonary nodules are benign. At follow-up, radiologists must also assess whether existing nodules are stable or growing and whether new nodules have appeared. This thesis investigated how artificial intelligence (AI) could support volume-based nodule management using data from the UK Lung Cancer Screening (UKLS) trial. A systematic review showed that lung cancer growth rates differ considerably by nodule type and histology, highlighting the importance of volume doubling time (VDT) in follow-up assessment. At baseline screening, AI as a first reader showed potential to rule out negative scans, with an estimated maximum reduction of 79% in radiologist CT-reading workload. Benign perifissural nodules (PFNs) contributed little additional workload although the AI could not classify them. At follow-up, AI identified rapid growth in confirmed lung cancers and automatically matched most persisting nodules across CT scans. Most matching failures involved non-nodular structures and had little clinical impact. Separately, the new-nodule study evaluated the NELSON-derived volume thresholds of 30 and 200 mm³ and found a higher malignancy risk for new solid nodules detected at 12 months than at three months. Together, these findings support the development of AI as a first reader at both baseline and follow-up screening. However, a complete AI-first follow-up workflow, including automated detection of new nodules, still requires further validation.

More information

The public defence will start exactly at 15.30 hrs. The doors will be closed once the public defence starts, latecomers can access the hall via the fourth floor. Given the solemn nature of the meeting, we advise not to bring children under the age of 6 to the first part of the ceremony.
A livestream link has been provided to candidate.

Compare 0 study programme