Deep Learning Time-Series Prediction of Long-Term Growth Patterns of Pulmonary Ground-Glass Nodules Using Serial CT

Condition(s) studied

Investigational drug(s) / intervention(s)

Serial Thin-Slice Chest CT: Routine-care thin-slice non-contrast chest CT (slice thickness ≤ 1.5 mm, lung-window reconstruction) acquired at baseline and at subsequent clinical follow-up timepoints (minimum inter-scan interval \> 1 month). Images are resampled to 1 × 1 × 1 mm and intensity-normalized before analysis. No additional imaging, radiation exposure, or procedures are performed for this study; all imaging is part of routine clinical care.

Study summary

Pulmonary ground-glass nodules (GGNs) are commonly found on chest CT scans. Some stay stable for years, while others slowly or rapidly turn into lung cancer. Doctors currently follow these nodules with repeated CT scans, but it is difficult to tell ahead of time which nodules will progress, how fast they will progress, and which ones can be safely monitored rather than immediately treated.

This observational study aims to develop and validate an artificial intelligence (AI) model that uses each patient's series of CT scans over time to predict the long-term growth behavior of a GGN. The research team will collect three retrospective single-center cohorts from Peking University People's Hospital (a development cohort and two internal test cohorts, one from surgically resected patients and one from non-operated patients followed by serial CT) as well as a prospective multi-center validation cohort enrolled after the AI model is locked.

For every patient, each GGN is automatically segmented in three dimensions on every CT scan. A deep learning model extracts imaging features at each timepoint and feeds the sequence of features, together with the actual times between scans, into a time-aware sequence model. The model is trained to predict (i) whether the nodule will show radiological progression at 1, 3, and 5 years after baseline, and (ii) which of four long-term growth patterns the nodule will follow: stable, slow progression, slow-then-rapid progression, or rapid progression. In patients who were ultimately resected, the histopathological diagnosis serves as a secondary reference standard.

This is an observational study. No experimental treatment is given. All CT scans and clinical visits are part of routine clinical care.

Eligibility

Primary outcome measure(s)

• Radiological Progression of the Target Ground-Glass Nodule at 1 Year — 12 months from baseline CT (± 3-month window)
  Radiological progression is defined as meeting either of the following on a follow-up thin-slice CT compared with the baseline CT, based on 3D automated segmentation with expert adjudication: (a) increase of the overall maximum diameter of the nodule by ≥ 2 mm; OR (b) the appearance of a new solid component, or the increase in maximum diameter of an existing solid component, by ≥ 2 mm. The solid component is defined as regions with an attenuation value greater than -300 HU. Each participant is classified as a progression event at the 1-year timepoint if either criterion is met on a CT performed within the ± 3-month window around 12 months after baseline. Applies to all four cohorts.
• Radiological Progression of the Target Ground-Glass Nodule at 3 Years — 36 months from baseline CT (± 6-month window)
  Same progression definition as Primary Outcome 1, assessed on a CT performed within the ± 6-month window around 36 months after baseline. Ascertained where the available follow-up duration permits. Applies to all four cohorts.
• Radiological Progression of the Target Ground-Glass Nodule at 5 Years — 60 months from baseline CT (± 6-month window)
  Same progression definition as Primary Outcome 1, assessed on a CT performed within the ± 6-month window around 60 months after baseline. Ascertained where the available follow-up duration permits. Applies to all four cohorts.
• Long-Term Growth Trajectory Classification of the Target GGN — Assessed across the full serial CT record, up to 60 months from baseline
  Each participant's target GGN is assigned, based on the full serial CT record, to exactly one of four mutually exclusive trajectory classes: (1) Stable - no progression event during follow-up; (2) Slow progression - continuous, approximately constant slow growth or slow increase of the solid component; (3) Slow-then-rapid progression - stable or minimally changing for an early period (e.g., 1-3 years) followed by abrupt acceleration (e.g., marked shortening of volume doubling time or new prominent solid component); (4) Rapid progression - aggressive growth evident early in follow-up. Classification is performed by two senior chest radiologists reading independently on the 3D automated outputs, with a third senior radiologist adjudicating disagreements.

Trial sites (1)

More Peking University People's Hospital trials in China

Other trials for the same condition