Biology subjects

Lam, P.

Publications and source records attributed to Lam, P..

3 recordsLinked to original sources

bioRxiv2021

Lamina cribrosa vessel and collagen beam networks are distinct

Our goal was to analyze the spatial interrelation between vascular and collagen networks in the lamina cribrosa (LC). Specifically, we quantified the percentages of collagen beams with/without vessels and of vessels inside/outside of collagen beams. To do this, the vasculature of six normal monkey eyes was labelled by perfusion post-mortem. After enucleation, coronal cryosections through the LC were imaged using fluorescence and polarized light microscopy to visualize the blood vessels and collagen beams, respectively. The images were registered to form 3D volumes. Beams and vessels were segmented, and their spatial interrelationship was quantified in 3D. We found that 22% of the beams contained a vessel (range 14% to 32%), and 21% of vessels were outside beams (13% to 36%). Stated differently, 78% of beams did not contain a vessel (68% to 86%), and 79% of vessels were inside a beam (64% to 87%). Individual monkeys differed significantly in the fraction of vessels outside beams (p<0.01 by linear mixed effect analysis), but not in the fraction of beams with vessels (p>0.05). There were no significant differences between contralateral eyes in the percent of beams with vessels and of vessels outside beams (p>0.05). Our results show that the vascular and collagenous networks of the LC in monkey are clearly distinct, and the historical notions that each LC beam contains a vessel and all vessels are within beams are inaccurate. We postulate that vessels outside beams may be relatively more vulnerable to mechanical compression by elevated IOP than are vessels shielded inside of beams. Research highlights- We combined fluorescence and polarized light microscopy to map in 3D the lamina cribrosa vessels and collagen beams of three pairs of monkey eyes - Collagen beam and vessel networks of the lamina cribrosa have distinct topologies - Over half of lamina cribrosa collagen beams did not contain a blood vessel - One fifth of blood vessels in the lamina cribrosa were outside collagen beams - Beams with/without vessels and vessels inside/outside beams may respond differently to IOP

bioengineering↗

bioRxiv2021

Alzheimer's Disease Classification Accuracy is Improved by MRI Harmonization based on Attention-Guided Generative Adversarial Networks

Alzheimers disease (AD) accounts for 60% of dementia cases worldwide; patients with the disease typically suffer from irreversible memory loss and progressive decline in multiple cognitive domains. With brain imaging techniques such as magnetic resonance imaging (MRI), microscopic brain changes are detectable even before abnormal memory loss is detected clinically. Patterns of brain atrophy can be measured using MRI, which gives us an opportunity to facilitate AD detection using image classification techniques. Even so, MRI scanning protocols and scanners differ across studies. The resulting differences in image contrast and signal to noise make it important to train and test classification models on multiple datasets, and to handle shifts in image characteristics across protocols (also known as domain transfer or domain adaptation). Here, we examined whether adversarial domain adaptation can boost the performance of a Convolutional Neural Network (CNN) model designed to classify AD. To test this, we used an Attention-Guided Generative Adversarial Network (GAN) to harmonize images from three publicly available brain MRI datasets - ADNI, AIBL and OASIS - adjusting for scanner-dependent effects. Our AG-GAN optimized a joint objective function that included attention loss, pixel loss, cycle-consistency loss and adversarial loss; the model was trained bidirectionally in an end-to-end fashion. For AD classification, we adapted the popular 2D AlexNet CNN to handle 3D images. Classification based on harmonized MR images significantly outperformed classification based on the three datasets in non-harmonized form, motivating further work on image harmonization using adversarial techniques.

neuroscience↗

bioRxiv2020

Spatial analysis of ligand-receptor interactions in skin cancer at genome-wide and single-cell resolution

The ability to study cancer-immune cell communication across the whole tumor section without tissue dissociation is needed for cancer immunotherapies, to understand molecular mechanisms and to discover potential druggable targets. In this work, we developed a powerful experimental and analytical toolbox to enable genome-wide scale discovery and targeted validation of cellular communication. We assessed the utilities of five sequencing and imaging technologies to study cancer tissue, including single-cell RNA sequencing and Spatial Transcriptomic (measuring over >20,000 genes), RNA In Situ Hybridization (multiplex 4-12 genes), digital droplet PCR, and Opal multiplex protein staining (4-9 proteins). To spatially integrate multimodal data, we developed a computational method called STRISH that can automatically scan across the whole tissue section for local expression of gene and/or protein markers to recapitulate an interaction landscape across the whole tissue. We evaluated the unique ability of this toolbox to discover and validate cell-cell interaction in situ through in-depth analysis of two types of cancer, basal cell carcinoma and squamous cell carcinoma, which account for over 70% of cancer cases. We expect that the approach described here will be widely applied to discover and validate ligand receptor interaction in different types of solid cancer tumors.

cancer biology↗