Early Prediction of Pathological Complete Response to Neoadjuvant Chemotherapy Using Temporal Deep Learning on DWI
Authors: Pablo García Marcos, Md. Tarequl Islam, Paula Puerta González, Guillermo Lorenzo, Héctor Gómez, Covadonga del Camino, Angel Rio-Alvarez, Víctor M. González
Organizations: Computer Sciences Department and the Biomedical Engineering Center (BME), University of Oviedo, 33003 Asturias, Spain · Rabindra University, Sirajganj-6770, Bangladesh · Electrical Engineering Department and the Biomedical Engineering Center (BME), University of Oviedo, 33003 Asturias, Spain · Group of Numerical Methods in Engineering, Department of Mathematics, University of A Coruña, 15071 A Coruña, Spain · Oden Institute for Computational Engineering and Sciences, The University of Texas at Austin, Austin, TX 78712 USA · School of Mechanical Engineering, the Weldon School of Biomedical Engineering, and the Purdue Center for Cancer Research, Purdue University, West Lafayette, IN 47907 USA · Radiodiagnostic Service, Asturias Central University Hospital (HUCA), 33011 Oviedo, Spain
Early identification of non-responders to neoadjuvant chemotherapy (NACT) is crucial for timely treatment adaptation in breast cancer. However, many existing predictive models rely on multiparametric magnetic resonance imaging (MRI), late treatment time points, or extensive clinical data, which limits their applicability. This study proposes a deep learning framework for early prediction of pathological complete response (pCR) using only diffusion-weighted MRI (DW-MRI) acquired at baseline and after the first NACT cycle. This framework feeds cropped tumor-centered patches to an EfficientNet-based temporal model that directly learns tumor shape and local tissue characteristics without explicit radiomic feature engineering. The model, trained with 10-fold cross-validation, achieved an area under the receiver operating characteristic curve (AUC) of 0.90 for pCR prediction after one cycle, providing actionable information after a single treatment cycle while avoiding gadolinium administration and reducing dependence on heterogeneous clinical data. By focusing on the baseline-to-first-cycle window instead of later stages, the approach supports earlier escalation or de-escalation of NACT, and its exclusive reliance on DW-MRI facilitates protocol standardization, multi-centre deployment and privacy-preserving data sharing. These results demonstrate that DW-MRI-based deep learning on tumor-centered patches constitutes a minimally invasive, clinically deployable strategy for early pCR prediction, with direct implications for personalized treatment adaptation in neoadjuvant breast cancer therapy.
Figures & tables
Fig. 1 : Slice of a study of the dataset in DW (left), ADC (center) and mask (right) formats.
Fig. 2 : Pre-processing pipeline. Values of the arrows represent Patients×Time points×Slices×Width×Height .
Fig. 3 : NACT response prediction model architecture.
Backbone
Temporal
Acc (%)
F1
EfficientNet-B0
LSTM
76.0
0.77
EfficientNet-B0
Non-recurrent
75.0
0.76
EffNet-B3 (RA)
LSTM
74.0
0.74
EffNet-B3 (RA)
Non-recurrent
71.0
0.72
EffNetV2-B3
LSTM
72.0
0.72
EffNetV2-B3
Non-recurrent
72.0
0.72
TABLE I : Backbone Ablation on the T0–T1 Setting Using 5-Fold Cross-Validation on 50×50 Tumor Patches. Mean Validation Accuracy and F1-Score Across Folds for Different Backbones and Temporal Modules (No Quality Indices).
LSTM configuration
Acc (%)
F1
32 units, dropout 0.30
74.0
0.74
32 units, dropout 0.40
75.0
0.76
64 units, dropout 0.30
75.0
0.76
64 units, dropout 0.40
76.0
0.77
TABLE II : Ablation on LSTM Hidden Size and Dropout Rate for the EfficientNet-B0 Backbone in the T0–T1 Setting Using 5-Fold Cross-Validation on 50×50 Tumor Patches.
Feature set
Acc (%)
F1
Imaging features only
76.0
0.77
Imaging + quality indices
77.0
0.77
TABLE III : Effect of Adding Image Quality Indices (QC) to the Imaging Features Extracted from 50×50 Tumor-Centered Patches (T0–T1, EfficientNet-B0 + LSTM), Using 5-Fold Cross-Validation.
Fold
Acc (%)
F1
AUC
1
88.24
0.88
0.92
2
94.12
0.94
0.97
3
88.24
0.88
0.86
4
70.59
0.71
0.76
5
88.24
0.88
0.91
6
82.35
0.83
0.94
TABLE IV : Per-Fold Validation Accuracy (Acc), F1-Score and AUC for the Final T0–T1 Model with Image Quality Indices (10-Fold Cross-Validation, 50×50 Tumor-Centered Patches).
Fig. 4 : ROC curve showing the average performance across the 10 cross-validation folds.
TABLE V : Comparison of Previous Studies and the Proposed Approach for Early Prediction of pCR to NACT in Breast Cancer. Only MRI-Related Metrics Are Reported for Papers Employing Multiple Approaches.
Pathologic complete response (pCR) is an important endpoint in neoadjuvant chemotherapy (NAC) for breast cancer, and predicting pCR from imaging during treatment could support treatment response assessment. Many existing imaging-based approaches rely on a single static timepoint, which fails to capture changes that occur during treatment. In this work, we present a longitudinal framework that combines a frozen 3D foundation encoder (Pillar-0) with our Temporal Dynamics Network (TDN) to predict treatment response from serial Dynamic Contrast-Enhanced (DCE) MRI acquired across four clinical timepoints from pre-treatment to pre-surgery. The TDN combines time-aware volumetric embeddings with clinical and treatment data to predict pCR. Evaluated on 982 patients from the combined I-SPY2 and ACRIN-6698 cohort, the proposed model achieves strong performance across all reported metrics when longitudinal 3D imaging is fused with clinical data (test AUROC: 73.6%, balanced accuracy: 69.1%). While clinical variables provide the strongest individual predictive signal, longitudinal 3D imaging contributes complementary information when fused with clinical data, improving pCR prediction. Our source code is available at: https://github.com/omarftt/longitudinal_temporal_pillar.
Fidel Omar Tito Cruz, Neda Ghafouri, Zengyan Wang +3
Prediction of pathological complete response before neoadjuvant chemotherapy may facilitate more tailored therapeutic planning for breast cancer patients. This work proposes a deep-learning model for pretreatment data only, combining apparent diffusion coefficient maps, dynamic contrast-enhanced magnetic resonance imaging, and clinical variables. The study uses the public ACRIN 6698/I-SPY2 multicenter dataset. The architecture employs EfficientNet-B0 pretrained encoders for image feature extraction and late fusion with clinical information. Multiple clinical variables were evaluated, including age, race, histological type, HR/HER2 subtype, SBR grade, and maximum diameter. Only HR/HER2 subtype improved the average area under the receiver operating characteristic curve (AUC) and was retained in the final model. Using stratified five-fold cross-validation, standalone apparent diffusion coefficient maps achieved a mean AUC of 0.79, whereas dynamic contrast-enhanced magnetic resonance imaging achieved 0.74. Adding HR/HER2 subtype improved performance to 0.83 and 0.81, respectively. The final configuration, using both imaging modalities and HR/HER2 subtype, achieved an AUC of 0.86. These results support pretreatment multimodal learning for response prediction, although external validation is required before clinical use.
Pablo García Marcos, Paula Puerta González, Guillermo Lorenzo +6
Computer Science Department, University of Oviedo, Asturias, Spain · Biomedical Engineering Center (BME), University of Oviedo, Asturias, Spain · Electrical Engineering Department, University of Oviedo, Asturias, Spain +7
Neoadjuvant chemotherapy (NAC) response prediction is clinically important for treatment stratification in breast cancer. However, robust pre-treatment pathological complete response (pCR) prediction remains challenging due to insufficient cross-modal modeling, multicenter imaging heterogeneity, and weak evidence-grounded interpretability. We propose ClinRAG-GRAPH, a Clinically informed Retrieval-Augmented Generation Graph framework, for pre-treatment pCR prediction from DCE-MRI, structured clinical variables, and biopsy-derived pathological biomarkers. ClinRAG-GRAPH constructs an intra-patient clinical-prior graph and applies a prior-guided relation-aware graph convolutional network for structured multimodal representation learning. To improve cross-center robustness, we introduce a dual-branch domain-adversarial learning strategy to suppress protocol-related MRI bias while preserving pCR-relevant features. To enhance interpretability, we further incorporate large language model (LLM)-driven subgraph RAG module that retrieves clinically analogous historical cases and integrates retrieved evidence for pCR inference. We assemble a large-scale multicenter NAC breast cancer cohort for extensive validation, drawing from two public sources and three in-house centers.Results show that ClinRAG-GRAPH achieves AUCs of 0.815 on the internal test set and 0.774/0.712 on two external test sets, demonstrating robust pre-treatment pCR prediction across centers. The code is available at the anonymized https://github.com/miccai26-1181/ClinRAG-GRAPH.
Yaofei Duan, Yuhao Huang, Tianyu Zhang +12
Department of Medical Imaging, Radboud University Medical Center, Nijmegen, The Netherlands · The Netherlands Cancer Institute, Amsterdam, The Netherlands · Faculty of Applied Sciences, Macao Polytechnic University, Macau, China +5