Deep learning workflow to support in-flight processing of digital aerial imagery for wildlife population surveys

Deep learning shows promise for automating detection and classification of wildlife from digital aerial imagery to support cost-efficient remote sensing solutions for wildlife population monitoring. To support in-flight orthorectification and machine learning processing to detect and classify wildlife from imagery in near real-time, we evaluated deep learning methods that address hardware limitations and the need for processing efficiencies to support the envisioned in-flight workflow. We developed an annotated dataset for a suite of marine birds from high-resolution digital aerial imagery collected over open water environments to train the models. The proposed 3-stage workflow for automated, in-flight data processing includes: 1) image filtering based on the probability of any bird occurrence, 2) bird instance detection, and 3) bird instance classification. For image filtering, we compared the performance of a binary classifier with Mask Region-based Convolutional Neural Network (Mask R-CNN) as a means of sub-setting large volumes of imagery based on the probability of at least one bird occurrence in an image. On both the validation and test datasets, the binary classifier achieved higher performance than Mask R-CNN for predicting bird occurrence at the image-level. We recommend the binary classifier over Mask R-CNN for workflow first-stage filtering. For bird instance detection, we leveraged Mask R-CNN as our detection framework and proposed an iterative refinement method to bootstrap our predicted detections from loose ground-truth annotations. We also discuss future work to address the taxonomic classification phase of the envisioned workflow.

Neoantigen Cancer Vaccine for Immunologically Cold Microsatellite-stable Colorectal Cancer

PURPOSE/OBJECTIVE(S)

Immunotherapies, such as immune checkpoint inhibitors (ICIs), have revolutionized management of some cancers but have little benefit for microsatellite-stable colorectal cancer patients (MSS-CRC). This is, in part, due to the low mutations and neoantigen expression in this immunogenically "cold" MSS-CRC. Therefore, we aim to develop novel shared neoantigen-based therapeutic cancer vaccine to reinvigorate antitumor immunity and enhance the therapeutic benefit of radiotherapy in MSS-CRC.

MATERIALS/METHODS

To identify novel highly expressed and shared neoantigens, we collected 40 match-paired adjacent normal and tumor tissues from MSS-CRC patients for WES-seq, RNA-seq, and liquid chromatography-MS/MS (LC-MS/MS). By incorporating these databases, we established Neoantigen Discovery and Validation (NeoDiva) system to identify a cluster of highly expressed and shared neoantigens derived from non-coding regions and evaluate its immunogenicity by HLA-A*11 transgenic mice. We then develop a neoantigen-based therapeutic cancer vaccine by an engineered adenovirus-associated virus (AAV) to evaluate its therapeutic efficacy in combination with radiotherapy in MSS-CRC animal model.

RESULTS

We identified a cluster of highly expressed and shared neoantigens (HLA-A*11-restricted) derived from non-coding regions. The immunogenicity of these novel neoantigens was demonstrated by HLA-A*11 transgenic mice and ex vivo stimulation. Moreover, the engineered AAV-based neoantigen cancer vaccine significantly eradicates cancer cells, prevents distant metastasis, prolong survival period in combination with radiotherapy. By flow cytometry, ELISPOT and MHC-I-tetramer assay, we demonstrated the recruitment of tumor-infiltrating lymphocytes was remarkably increased and neoantigen-specific T cell response was enhanced. Moreover, these isolated neoantigen-specific T cells can recognize cancer cells and produce IFNg to kill cancer cells.

CONCLUSION

Neoantigens identified by our NeoDiVa platform, via the combination of radiotherapy and a novel AAV vaccine delivery system, boosted antigen-specific T-cell function and improve tumor control of limnologically "cold" MSS colorectal cancer in vivo. We are in the process of obtaining an IND and initiating Phase I/II clinical trial to validate safety and efficacy of these exciting findings.

A convolutional neural network-based screening tool for X-ray serial crystallography

A new tool is introduced for screening macromolecular X-ray crystallography diffraction images produced at an X-ray free-electron laser light source. Based on a data-driven deep learning approach, the proposed tool executes a convolutional neural network to detect Bragg spots. Automatic image processing algorithms described can enable the classification of large data sets, acquired under realistic conditions consisting of noisy data with experimental artifacts. Outcomes are compared for different data regimes, including samples from multiple instruments and differing amounts of training data for neural network optimization.