Computer vision and machine‐learning techniques for quantification and predictive modeling of intracellular anticancer drug delivery by nanocarriers. Issue 1 (18th December 2021)

• Record Type: Journal Article
• Authors: Goswami, Sanjay; Dhobale, Kshama D.; Wavhale, Ravindra D.; Goswami, Barnali; Banerjee, Shashwat S.
• Is Part Of: Applied AI Letters. Volume 3:Issue 1(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12-18
• Subjects: bio‐image analysis -- cancer -- computer vision -- machine learning -- nanocarriers -- predictive modeling; 006.3
• DOI: 10.1002/ail2.50
• Languages: English
• ISSNs: 2689-5595

Abstract

Abstract: The field of cancer nanomedicine has made significant progress, but its clinical translation is impeded by many challenges, such as the difficulty in analyzing intracellular anticancer drug release by the nanocarriers due to the lack of suitable tools. Here, we propose the development of an image‐based strategy involving machine learning (ML) to evaluate anticancer drug such as doxorubicin hydrochloride (DOX) released by a nanocarrier inside the HCT116 colon cancer cells and its subsequent intracellular accumulation. This technique combines fluorescent cell imaging with ML‐based image analysis to assess and quantify the delivery of DOX by nanoparticles within them. We show that DOX in HCT116 cells was higher for multifunctional CNT‐DOX‐Fe3 O4 nanocarrier than free DOX, indicating efficient and steady release of DOX as well as superior retentive property of the nanocarrier. Initially (1 and 4 hours), the luminance intensity of DOX in the cell cytoplasm delivered by CNT‐DOX‐Fe3 O4 nanocarrier was ~0.34 and ~0.42 times lesser than that of free DOX delivered normally. However, at 24 and 48 hours posttreatment, the luminance intensity of DOX for CNT‐DOX‐Fe3 O4 nanocarrier was ~1.98 and ~1.92 times higher than that of free DOX. Furthermore, the luminance intensity of DOX for CNT‐DOX‐Fe3 O4 in the whole cell was ~1.35 and ~1.62 times higher than that of free DOX at 24 and 48 hours, respectively. The high‐throughput nature of our image analysis workflow allowed us to automate the process of DOX retention analysis and enabled us to devise ML‐based modeling to predict the percentage of anticancer drug retention in cells. The development of models to automatically quantify and predict intracellular drug release in cancer cells could benefit personalized treatments by optimizing the design of nanocarriers. Abstract : Four cell luminance images of DOX treated cancer cells have been utilized to draw a generalized machine learning model that can predict approximate accumulation of anti‐cancer drug delivered through nanocarriers.