A supercontinuum source based on a 1-mm ² Si ₃ N ₄ chip offers improved OCT imaging performance over commercial supercontinuum sources.

Optical coherence tomography (OCT) is being investigated in breast cancer diagnostics as a real-time histology evaluation tool. We present a customized deep convolutional neural network (CNN) for classification of breast tissues in OCT B-scans. Images of human breast samples from mastectomies and breast reductions were acquired using a custom ultrahigh-resolution OCT system with 2.72 µm axial resolution and 5.52 µm lateral resolution. The network achieved 96.7% accuracy, 92% sensitivity, and 99.7% specificity on a dataset of 23 patients. The usage of deep learning will be important for the practical integration of OCT into clinical practice.

Technologies allowing for real-time histology evaluation may impact the entire anatomic diagnostic process across multiple organs—from biopsy to intraoperative margin assessment to tissue grossing as well as tissue banking.^1,2 The macroscopic examination of excised tissue (i.e., tissue grossing) is a critical step in the diagnostic anatomic pathology process as it defines regions of interest requiring further analysis, but it is limited by the variable correlation between gross and microscopic findings.^3,4 This fundamental discordance is a gap that could be closed by the application of real-time histology evaluation by optical coherence tomography (OCT). OCT is a nondestructive, 3D microscopic imaging modality with cellular-level resolution⁵ that is being applied currently in clinical ophthalmology⁶ and dermatology⁷ and is emerging in other specialties.^8,9 OCT provides a resolution and depth of field that is midway between traditional microscopy and clinical ultrasound. Although ultrasound can achieve a large imaging depth (several centimeters), it has limited resolution ($>10\mu \mathrm{m}$). On the other hand, traditional microscopy techniques such as confocal microscopy have very high resolution ($<1\mu \mathrm{m}$), but the penetration depth is minimal ($<1\mathrm{mm}$). OCT bridges the gap between clinical and microscopic imaging, providing 1 to 2 mm of penetration depth and high resolution (1 to $10\mu \mathrm{m}$). Once integrated into the diagnostic workflow, we envision OCT improving the diagnostic process for radiologists by real-time assessment of biopsies; for surgeons by intraoperative margin determination; and for pathologists by identification of regions of interest in pathology grossing/tissue banking, increasing efficiencies of the laboratory by reducing block submission, and reducing pathologist workload by reviewing fewer slides. The aim of OCT in the diagnostic arena is to provide for real-time broad-spectrum assessment of tissue characterization including normal, abnormal, and neoplastic tissues.