Hyperspectral Imaging for Breast-Conserving Surgery Margin Assessment: A Systematic Review.

CONTEXT.—: Breast-conserving surgery (BCS) offers comparable overall survival rates to mastectomy. However, 14% to 37% of cases require re-excision owing to tumor-positive margins after BCS. Hyperspectral imaging (HSI) provides rapid, real-time, thorough margin assessment, reducing the burden of re-excisions. OBJECTIVE.—: To systematically review studies published up to January 2026 investigating HSI or multispectral imaging for breast cancer margin assessment. DATA SOURCES.—: Twenty-four primary studies involving HSI or multispectral imaging of fresh lumpectomy specimens or histologic slides in breast cancer were included, focusing on reporting methodology, imaging setup, machine learning models, diagnostic performance, and comparisons to standard techniques. Key data on sample types, imaging acquisition (spectral range, modality), preprocessing, algorithms (conventional classifiers and deep learning), validation methods, and diagnostic accuracy (sensitivity, specificity, and area under the receiver operating characteristic curve [AUC]) were extracted. CONCLUSIONS.—: HSI consistently differentiated malignant from normal breast tissue across studies. In ex vivo lumpectomy specimens, HSI classifiers achieved sensitivity of 83% to 96% and specificity of 72% to 96% for margin-level detection, with AUC approximating 0.9. Machine learning ranged from linear discriminant analysis and support vector machines to convolutional neural networks and clustering algorithms, leveraging visible spectrum to near-infrared spectral range. HSI was integrated into simulated surgical workflows, imaging a lumpectomy surface within ∼10 minutes and providing automated margin maps in 3 minutes. Bulky hardware, calibration requirements, reduced spectral resolution, and regulatory approval remain major challenges to clinical adoption. No HSI devices have regulatory approval for intraoperative margin assessment in BCS. HSI enables whole-surface visualization of tumor margins. Translating HSI requires multidisciplinary efforts and prospective multisite trials.