# LungGANDetectAI: a GAN-augmented and attention-guided deep learning framework for accurate and explainable lung cancer detection

> **Reliability concern:** This paper was retracted on 2026-03-17. See notice: 10.1038/s41598-026-44623-0.

- **DOI:** [10.1038/s41598-025-34750-5](https://doi.org/10.1038/s41598-025-34750-5)
- **Rating:** Limited (2.8 out of 5)
- **Verdict:** A plausible lung-CT CAD prototype combining GAN augmentation and attention, constrained by single-dataset slice-level validation.
- **Framework version:** v5.5
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-07-22
- **Evaluation page:** https://www.nabu.science/paper/10.1038%2Fs41598-025-34750-5

## Key claims

1. LungGANDetectAI achieves 98.4%, 98.2%, 98.3%, and 98.2% accuracy, precision, recall, and F1-score, respectively, only in malignant cases
   - Support: Tentative
   - Type: Primary finding
   - Basis: single small public dataset, selective malignant-only framing, and no external validation
   - Location: Abstract
2. LungGANDetectAI integrates class-balanced generative adversarial network (GAN) synthesis, attention-guided classification, and explainable visualisation
   - Support: Supported
   - Type: Secondary
   - Basis: architecture and methods describe all three modules, though novelty is incremental
   - Location: Introduction; Materials and methods, LungGANDetectAI framework
3. The results indicated consistent improvements in classification accuracy and recall (across all classes, especially those with low representation) in the presence of synthetic data
   - Support: Tentative
   - Type: Secondary
   - Basis: ablation is described qualitatively without complete quantitative reporting in the supplied text
   - Location: Materials and methods, Synthetic data contribution and utility
4. Grad-CAM is applied to gradually reveal the clinical interpretability of the network’s decision-making
   - Support: Tentative
   - Type: Secondary
   - Basis: visual explanations are provided, but no radiologist agreement or lesion-ground-truth validation is shown
   - Location: Abstract; Explainability and clinical interpretability integration
5. LungGANDetectAI is an accurate and interpretable decision-support tool that may serve as an early diagnostic aid for radiologists
   - Support: Unsupported
   - Type: Secondary
   - Basis: clinical-aid framing outruns retrospective slice-level benchmark evidence
   - Location: Abstract

## Strengths

- **Patient-level split reduces leakage risk** The Methods state that all slices from a given patient are assigned to the same training, validation, or test subset. This is a meaningful safeguard for slice-based CT classification, where random slice-level splitting can overstate performance.
  - Dimension: methodological rigour
  - Location: Materials and methods, Dataset description and experimental splits
- **Integrated augmentation-attention-explainability pipeline** The framework links GAN-based class balancing, a ResNet50 classifier with spatial and channel attention modules, and Grad-CAM visualizations. This gives the paper a coherent engineering objective even if the components are not individually novel.
  - Dimension: contribution
  - Location: Introduction; Materials and methods, LungGANDetectAI framework
- **Relevant clinical problem and users** The article targets early lung cancer detection from CT scans and frames the system as radiology decision support. The problem is concrete, active, and connected to identifiable clinical stakeholders.
  - Dimension: application pathway
  - Location: Abstract; Introduction

## Limitations

- **Single-dataset validation limits generalization** All reported performance is based on IQ-OTH/NCCD, approximately 1190 slices from 110 patients, with no external validation. This sharply limits confidence that the reported malignant-case metrics would hold across scanners, institutions, or clinical populations.
  - Dimension: methodological rigour
  - Location: Materials and methods, Dataset description and experimental splits
- **FID reporting indicates methodological imprecision** The Synthetic data contribution section states that FID greater than 40 would indicate generated images are usable and realistic. Because lower FID conventionally indicates closer real-synthetic distributional similarity, this undermines confidence in the GAN-quality assessment.
  - Dimension: methodological rigour
  - Location: Materials and methods, Synthetic data contribution and utility
- **Manuscript coherence and reporting problems** The Abstract emphasizes metrics “only in malignant cases” while the model is a three-class classifier, and the Methods contain inconsistent training descriptions and malformed notation. Garbled prose in the explainability section further reduces confidence in reporting precision.
  - Dimension: reporting
  - Location: Abstract; Table 2; Eq. 6; Explainability and clinical interpretability integration

## Reliability checks

Concerns (4 of 4 checks).

- **Internal coherence:** Noted. The manuscript is broadly followable but contains material reporting and coherence issues that reduce confidence in the precision of the methods and claims. These concerns warrant attention but do not by themselves establish that the core results are invalid.
- **Research conduct:** Clean. The study appears to use a named publicly available CT dataset and no prospective intervention is reported. Missing code, missing COI text, and lack of preregistration are recorded as openness or reporting limitations rather than conduct concerns for this design.
- **Reference integrity:** Clean. 93 references verified via PubMed, Crossref, or OpenAlex.
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Concern
- **Summary:** This paper was retracted on 2026-03-17. See notice: 10.1038/s41598-026-44623-0.
- **Last checked:** 2026-10-17