Circulating tumor cell detection in cancer patients using in-flow deep learning holography
Authors:
Kevin Mallery,
Nathaniel R. Bristow,
Nicholas Heller,
Yash Travadi,
Ali Arafa,
Kaylee Kamalanathan,
Catalina Galeano-Garces,
Mahdi Ahmadi,
Grant Schaap,
Alexa Hesch,
Olivia Hedeen,
Zikora Izuora,
Joel Hapke,
Jeffrey Miller,
Arjun Viswanathan,
Ivo Babris,
Songyi Bae,
Tuan Le,
Tony Clacko,
Emmanuel S. Antonarakis,
Badrinath R. Konety,
Justin M. Drake,
Jiarong Hong
Abstract:
Circulating tumor cells (CTCs) are cancer cells found in the bloodstream that serve as biomarkers for early cancer detection, prognostication, and disease monitoring. However, CTC detection remains challenging due to low cell abundance and heterogeneity. Digital holographic microscopy (DHM) offers a promising, label-free method for high-throughput CTC identification by capturing superior morpholog…
▽ More
Circulating tumor cells (CTCs) are cancer cells found in the bloodstream that serve as biomarkers for early cancer detection, prognostication, and disease monitoring. However, CTC detection remains challenging due to low cell abundance and heterogeneity. Digital holographic microscopy (DHM) offers a promising, label-free method for high-throughput CTC identification by capturing superior morphological information compared to traditional imaging methods, while remaining compatible with in-flow data acquisition. We present a streamlined DHM-based system that integrates microfluidic enrichment with deep learning-driven image analysis, supplemented by immunofluorescent profiling, to improve the sensitivity and specificity of CTC enumeration. Specifically, our platform combines inertial microfluidic preprocessing with dual-modality imaging, integrating holography with fluorescence sensing of up to two markers. A deep learning model, trained on a diverse set of healthy blood samples and cancer cell lines, and executed in real-time, provides a morphological confidence on a cell-by-cell basis that may then be combined with immunofluorescence criteria for enumeration. In a pilot study, we demonstrate significantly higher CTC counts in patients with late-stage prostate cancer (n=13) compared to healthy controls (n=8), with a patient-level false positive rate of 1 cell/mL. Notably, nearly two-thirds of identified CTCs were EpCAM-negative but PSMA positive (a prostate specific epithelial marker), suggesting that traditional use of EpCAM as an epithelial marker for CTCs may lead to false negatives. These findings highlight the potential of DHM for applications including but not limited to screening, diagnostics, and precision oncology.
△ Less
Submitted 9 July, 2025;
originally announced July 2025.
A Bayesian approach for energy-based estimation of acoustic aberrations in high intensity focused ultrasound treatment
Authors:
Bamdad Hosseini,
Charles Mougenot,
Samuel Pichardo,
Elodie Constanciel,
James M. Drake,
John M. Stockie
Abstract:
High intensity focused ultrasound is a non-invasive method for treatment of diseased tissue that uses a beam of ultrasound to generate heat within a small volume. A common challenge in application of this technique is that heterogeneity of the biological medium can defocus the ultrasound beam. Here we reduce the problem of refocusing the beam to the inverse problem of estimating the acoustic aberr…
▽ More
High intensity focused ultrasound is a non-invasive method for treatment of diseased tissue that uses a beam of ultrasound to generate heat within a small volume. A common challenge in application of this technique is that heterogeneity of the biological medium can defocus the ultrasound beam. Here we reduce the problem of refocusing the beam to the inverse problem of estimating the acoustic aberration due to the biological tissue from acoustic radiative force imaging data. We solve this inverse problem using a Bayesian framework with a hierarchical prior and solve the inverse problem using a Metropolis-within-Gibbs algorithm. The framework is tested using both synthetic and experimental datasets. We demonstrate that our approach has the ability to estimate the aberrations using small datasets, as little as 32 sonication tests, which can lead to significant speedup in the treatment process. Furthermore, our approach is compatible with a wide range of sonication tests and can be applied to other energy-based measurement techniques.
△ Less
Submitted 7 June, 2018; v1 submitted 24 February, 2016;
originally announced February 2016.
