EUVIMEDEUVIMED entdecken
Crossref

IGhost-BiLSTM with Explainable AI for Accurate Lung Cancer Classification and Prognosis from CT Volumes

Rajesh Tulasi, Dr. Vijaya Chandra Jadala

International Journal of Drug Delivery Technology · 08.05.2026 · 16 (25s)

DOI: 10.25258/ijddt.16.25s.74 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 08.05.2026. Verfasst von Rajesh Tulasi und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Worldwide, lung cancer remains one of the leading causes of death in cancer patients. For accurate CT scans, better diagnostic tools are needed because current methods often miss key details. Instead of relying on a single approach, this study combines two systems - one that filters noise from images and another that learns patterns over time. Called IGhost-BiLSTM, it uses a custom design called GhostNet to scan slices independently. Following each scan, a memorybased system tracks changes across frames. Because it splits tasks between layers, it handles motion and shape changes more smoothly. Decisions formed this way also show clearer reasoning when reviewed later. A key part of the design uses an intelligent system that clearly shows its decisions, mixing two methods - Gradient-weighted Class Activation Mapping and SHapley Additive exPlanations - to help doctors understand results. Tested thoroughly using standard images from LIDC-IDRI and LUNA16, it shows strong outcomes: accuracy near 96.7%, sensitivity at 99.1%, and tight calibration precision at 0.016. When mapped against actual radiology markings, the output lines up with 81.4% overlap on average. Experts in imaging rated its usefulness 4.71 out of 5. This approach runs faster, gives clearer results, and works well in medical settings - needing just under a quarter the number of components old systems used, yet still spotting diseases more accurately.

Zitieren

APA

Tulasi, R., & Jadala, D. V. C. (2026). IGhost-BiLSTM with Explainable AI for Accurate Lung Cancer Classification and Prognosis from CT Volumes. International Journal of Drug Delivery Technology, 16(25s). https://doi.org/10.25258/ijddt.16.25s.74

Vancouver

Tulasi R, Jadala DVC. IGhost-BiLSTM with Explainable AI for Accurate Lung Cancer Classification and Prognosis from CT Volumes. International Journal of Drug Delivery Technology. 2026;16(25s). doi:10.25258/ijddt.16.25s.74

BibTeX
@article{Tulasi2026,
  author = {Tulasi, Rajesh and Jadala, Dr. Vijaya Chandra},
  title = {IGhost-BiLSTM with Explainable AI for Accurate Lung Cancer Classification and Prognosis from CT Volumes},
  journal = {International Journal of Drug Delivery Technology},
  year = {2026},
  volume = {16},
  number = {25s},
  doi = {10.25258/ijddt.16.25s.74}
}
Kostenlos starten →