Virtual Library
Start Your Search
Calum Macaulay
Author of
-
+
MA15 - Usage of Computer and Molecular Analysis in Treatment Selection and Disease Prognostication (ID 141)
- Event: WCLC 2019
- Type: Mini Oral Session
- Track: Pathology
- Presentations: 1
- Now Available
- Moderators:John Le Quesne, Noriko Motoi
- Coordinates: 9/09/2019, 15:45 - 17:15, Tokyo (1982)
-
+
MA15.11 - Establishing a Cell Sociology Platform for the Assessment of Targetable Interactions to Predict Lung Cancer Outcome (Now Available) (ID 2652)
15:45 - 17:15 | Author(s): Calum Macaulay
- Abstract
- Presentation
Background
The tumor microenvironment (TME) is a complex mixture of tumor epithelium, stroma and immune cells. The immune component of the TME is highly prognostic for tumor progression and patient outcome. Immune functionality, however, is often dictated by direct cell-to-cell contacts and cannot be resolved by simple metrics of cell density (for example, number of cells per mm2 or flow cytometry). For example, direct contact between CD8+ T cells and target cells is necessary for CD8+ T cell activity, and direct contact between PD1+ and PD-L1+ cells is necessary for the efficacy of immune checkpoint inhibitors. Current immunohistochemistry (IHC) techniques identify immune cell numbers and densities, but lack assessment of spatial relationships (or “cell sociology”). Here, we develop a platform to examine these direct interactions within the TME, and assess their relationship with patient outcome in two independent non-small cell lung cancer (NSCLC) cohorts.
Method
Tissue sections of primary tumors from lung adenocarcinoma (LUAD) patients with known clinical outcome were stained using 2 multiplex IHC panels: CD3/CD8/CD79a (Panel 1) and PD1/PDL1/CD8 (Panel 2). Hyperspectral image analysis determined the phenotype of all cells. Using the same IHC panels, these observations were assessed in a secondary NSCLC dataset (n=674). Deconvolution of these images was used to identify cell types, and cellular ‘neighborhoods’ were assessed using a Voronoi approach. This cohort was also profiled by for gene expression to validate immune subset fractions. We further identified other tumor features, including the presence of tertiary lymphoid organs (TLOs; transient immune structures necessary for antibody production from B cells).
Result
High density of intra-tumoral CD8+ T cells was associated with non-recurrence of tumors. However, we find that a non-random cell sociology pattern of CD8+ T cells directly surrounded by tumor cells was more significantly associated with non-recurrence compared to density alone. Monte Carlo re‐sampling analysis determined that these cell sociology patterns were non-random.
Conclusion
Hyperspectral cell sociology expands our understanding of the complex interplay between tumor cells and immune infiltrate. This technology improves our understanding of the tumour microenvironment and allows us to directly quantify interactions that dictate immune responses to cancers. Consequently, the implementation of this platform could improve predictions of responses to immunotherapy and lead to a deeper understanding of anti-tumor immunity.
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
OA06 - Refining Lung Cancer Screening (ID 131)
- Event: WCLC 2019
- Type: Oral Session
- Track: Screening and Early Detection
- Presentations: 1
- Now Available
- Moderators:Tomasz Grodzki, Lluis Esteban Tejero
- Coordinates: 9/09/2019, 11:00 - 12:30, Hilton Head (1978)
-
+
OA06.01 - Comparison Between Radiomics-Based Machine Learning and Deep Learning Image Classification for Sub-Cm Lung Nodules (Now Available) (ID 2810)
11:00 - 12:30 | Presenting Author(s): Calum Macaulay
- Abstract
- Presentation
Background
New clinical challenges have arisen from the recent recognition for an improved mortality of cancers via lung cancer screening using LDCT. A particular challenge for physicians and CADx systems is the classification and prediction of behavior for sub-cm lung nodules that are frequently present in screening CT scans. By predicting and classifying the behavior of these small nodules, we can identify potential cancerous nodules into the earlier stages of malignancy making them more easily treatable.
Method
We have evaluated a multitude of image processing techniques to assist in CADx systems for these small nodules such as Radiomic feature-based machine learning algorithms (linear discriminant analysis) as well as leveraging pretrained convolution neural networks such as VGG19 and InceptionV3 using deep learning/transfer learning techniques. The linear discriminate Radiomic analysis (LDA) classified a sample of CT imaged nodules (n=514) using quasi-volumetric nodule data (images of the nodules from CT slices above and below the central slice) into three discriminate categories: cancerous (clinically confirmed, n = 140) versus resolved (not present in follow up CT scans, n=107) versus stable (a negligible change in shape, texture, size in multi year follow up CT scans, n=267). Each nodule was segmented from the original CT scan using an inhouse lung CT image segmenation routine. This routine generated 2167 discrete CT nodule images upon which 133 Radiomic shape and texture features were calculated.
Result
The LDA Radiomic analysis correctly classified the individual nodual sections with an accuracy of 75.1% (jackknife - leave one out result) using only 18 features predefined traditional image analysis features (4 shape feature(s), 14 texture feature(s)) for cancer vs resolved + stable nodules. Requiring that more than or equal to 50% of sections from a nodule be classified as cancer for the nodule to be classified as cancer individual nodules could be correctly classified with an 80% accuracy.
The leveraged pretrained networks (VGG19, and InceptionV3) trained using standard data augmentation and finetuning techniques, trained on this same quasi-volumetric image data as a binary classification task (malignant vs. benign nodules) achieved an average classification accuracy of 71% and 75% respectively through 10-crossfold validation.
Conclusion
Machine learning using 18 Radiomics features was able to classify 75.1% of the 2167 CT nodule images (up to 5 images/CT slices per nodule) and 80% of the nodules correctly. The best of the Deep Learning networks achieved almost equivalent results.
The image classification deep neural network results suggest the implementation of more advanced regularization and initialization deep learning techniques to further refine the decision boundaries for these pretrained networks might be benefitial. We believe the development of visualization neural network software to highlight the defining nodule features during classification would clinically assist in providing context clues for nodule diagnosis.
This work has been supported by TFRI project ref:1068
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
