Filtering by
- All Subjects: Drug Delivery
- All Subjects: deep learning
- Creators: Vernon, Brent
- Status: Published
Handheld, portable confocal laser endomicroscopy (CLE) is being explored in neurosurgery for its ability to image histopathological features of tissue at cellular resolution in real time during brain tumor surgery. Over the course of examination of the surgical tumor resection, hundreds to thousands of images may be collected. The high number of images requires significant time and storage load for subsequent reviewing, which motivated several research groups to employ deep convolutional neural networks (DCNNs) to improve its utility during surgery. DCNNs have proven to be useful in natural and medical image analysis tasks such as classification, object detection, and image segmentation.
This thesis proposes using DCNNs for analyzing CLE images of brain tumors. Particularly, it explores the practicality of DCNNs in three main tasks. First, off-the shelf DCNNs were used to classify images into diagnostic and non-diagnostic. Further experiments showed that both ensemble modeling and transfer learning improved the classifier’s accuracy in evaluating the diagnostic quality of new images at test stage. Second, a weakly-supervised learning pipeline was developed for localizing key features of diagnostic CLE images from gliomas. Third, image style transfer was used to improve the diagnostic quality of CLE images from glioma tumors by transforming the histology patterns in CLE images of fluorescein sodium-stained tissue into the ones in conventional hematoxylin and eosin-stained tissue slides.
These studies suggest that DCNNs are opted for analysis of CLE images. They may assist surgeons in sorting out the non-diagnostic images, highlighting the key regions and enhancing their appearance through pattern transformation in real time. With recent advances in deep learning such as generative adversarial networks and semi-supervised learning, new research directions need to be followed to discover more promises of DCNNs in CLE image analysis.
Carbohydrate counting has been shown to improve HbA1c levels for people with diabetes. However, the learning curve and inconvenience of carbohydrate counting make it difficult for patients to adhere to it. A deep learning model is proposed to identify food from an image, where it can help the user manage their carbohydrate counting. This early model has a 68.3% accuracy of identifying 101 different food classes. A more refined model in future work could be deployed into a mobile application to identify food the user is about to consume and log it for easier carbohydrate counting.
Alginate microspheres have recently become increasingly popular in the realm of drug delivery for their biocompatibility, nontoxicity, inexpensiveness, among other factors. Recent strict regulations on microsphere size have drastically increased manufacturing cost and waste, even though the effect of size variance on drug delivery and subsequent performance is unclear. If sphere size variance does not significantly affect drug release profiles, it is possible that future ordinances may loosen tolerances in manufacturing to limit waste produced and expenditures. We use a mathematical model developed by Nickel et al. [12], to theoretically predict drug delivery profiles based on sphere size, and correlate the expected release with experimental data. This model considers diffusion as the key component for drug delivery, which is defined by Fick’s Laws of Diffusion. Alginate, chosen for its simple fabrication method and biocompatibility, was formed into microspheres with a modified extrusion technique and characterized by size. Size variance was introduced in batches and delivery patterns were compared to control groups of identical size. Release patterns for brilliant blue dye, the mock drug chosen, were examined for both groups via UV spectrometry. The absorbance values were then converted to concentration value using a calibration curve done prior to experimentation. The concentration values were then converted to mass values. These values then produced curves representing the mass of the drug released over time. Although the control and experimental values were statistically significantly different, the curves were rather similar to each other. However, when compared to the predicted release pattern, the curves were not the same. Unexpected degradation caused this dissimilarity between the curves. The predictive model was then adjusted to account for degradation by changing the diffusion coefficient in the code to a reciprocal first order exponent. The similarity between the control and experimental curves can insinuate the notion that size tolerances for microsphere production can be somewhat lenient, as a batch containing fifteen beads of the same size and one with three different sizes yields similar release patterns.