Rapid inter-procedural 3D scanning of fine-needle aspiration cytology slides with a parallelized microscope - Significance: Fine needle aspiration (FNA) biopsies are a critical first step in the cancer diagnosis pipeline, wherein specimen material is extracted from a suspected pathological lesion and smeared onto a microscope slide. Cytopathologists then examine three-dimensional groups of cells across the entire slide for diagnostic purposes. Ideally, a cytopathologist is available in-person to perform “rapid on-site evaluation” (ROSE) during such FNA biopsy procedures, like bronchoscopies, which is known to increase positive predictive value and decreases procedure time for cancer diagnosis. Unfortunately, relatively few cytotechnologists and pathologists are available to perform ROSE, most notably outside of large hospital settings. Without ROSE, proceduralists generally do not know if they have obtained adequate specimen material for diagnostic review, or how to triage such samples, which leads to additional patient visits, additional procedures, and increased risk of side effect and harm. Proposal: In this project, Ramona Optics and the Duke Medical Center aim to develop a new imaging system, termed a multi-camera array scanner (MCAS), that rapidly digitizes and automatically analyzes thick FNA cytology smears in 3D. The MCAS will provide real-time (<1 min.) diagnostic feedback regarding specimen adequacy and diagnostic category (e.g., malignant or benign) during FNA biopsy procedures, such as to bronchoscopies, which can assist in scenarios where cytopathologists are not available for in-person specimen review. In this proposal, we aim to show how the MCAS’s automated inter-procedural analysis of FNA cytology can empower clinicians with valuable insights in real-time. Apart from empowering clinicians like bronchoscopists, we believe the MCAS can increase health equity, improve FNA biopsy procedure success, and help inform optimal next steps within the care pipeline. SA1 - Enhance MCAS scanning speed and algorithm accuracy: We will develop rapid focusing hardware, fast data transfer firmware and novel image screening software to ensure MCAS 3D scan and analysis speed is performed on whole-slide FNA cytology smears in < 1 min. for inter-procedural use. SA2 – Develop software for automated adequacy assessment and diagnostics: In collaboration with Duke Medical Center, we will scan 500 archival lung FNA cytology smears and train 3 unique machine learning algorithms to automatically assess 1) specimen adequacy, 2) malignancy, and 3) diagnostic category. Software will also display images of salient FNA specimen areas for a transparent review process. SA3 - Develop MCAS user interface and perform clinician-focused testing: In a final series of user tests with 6 clinicians at the Duke Medical Center, we aim to obtain direct feedback regarding MCAS hardware and software for automated and remote ROSE during simulated bronchoscopies, which will guide future development and clinical testing during bronchoscopy procedures within a planned Phase II project.
