Harvey Lui’s research while affiliated with BC Cancer Research Centre and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (305)


The volumetric multimodality microscopy imaging system and the method for volumetric data acquisition and reconstruction
a Schematic drawing of the volumetric multimodal microscopy optical imaging system based on vertical sectioning tissue imaging. A femtosecond (fs) laser beam after passing through a polarizing beamsplitter was scanned in the x-direction by a resonance scanning mirror and relayed to the back aperture of the objective. The focal point of the laser beam is scanned in the z-direction by moving the objective using a piezo positioner. Two-photon fluorescence (TPF) and second harmonic generation (SHG) signals are collected by the objective and reflected by the dichroic mirror and the beamsplitter to two photomultiplier tubes (PMTs). Reflectance confocal (RCM) signal is collected by the objective and reflected to the avalanche photodiode (APD) by the polarizing beamsplitter after passing through the dichroic mirror and the scanning mirrors. b The procedure for volumetric data acquisition and motion correction. Volumetric data was acquired by simultaneously scanning in the xz direction and (stretching) moving the skin with the translation stage in the y-direction. The acquired raw data is then reconstructed after skin surface detection, motion detection, motion correction, and frame averaging process (see details in the “Methods” section).
Illustration of the skin measurement interface and the vertical plane imaging results
a shows how the skin site is interfaced with the imaging head. b and c show two types of interface configurations. b is used for direct measurement of the skin, which includes a window adapter, a metal window, and a plastic plate with a hole in the center. c is used for motion-suppressed measurement, which includes a window adapter, a metal window, and a coverslip. d–g Example of xz vertical sectioning images acquired in vivo on a dorsal forearm of a 46-year-old male volunteer at the excitation wavelength of 785 nm and the merged pseudocolor images of the three channels show various skin layers. SC: stratum corneum, SG: stratum granulosum, SS: stratum spinosum, SB: stratum basale (marked by arrows), DP: dermal papilla. The white rectangular in (d) outlines a multilayer sandwich-like skin structure near the skin surface, the dark layer is assumed to be stratum lucidum (SL). White arrows point to basale cells containing melanin. Color is coded in green for RCM, red for TPF, and magenta for SHG. Scale bar: 50 µm.
Extended field of view (200 µm × 200 µm × 3.2 mm) volumetric imaging (raw data) from the dorsal forearm of a 29-year-old male volunteer with an excitation wavelength of 880 nm
The top row shows the three raw videos acquired from the three modalities: RCM (a), TPF (b), and SHG (c). Each video has 4072 frames. The bottom row (d) shows the reconstructed 3D image of the imaged skin site before motion correction. Scale bar: 50 µm.
Motion-correction algorithm
a The image of the raw volume (RCM) with motion artifacts. b The method to calculate the relative motion between every two neighbor frames. b-1 and b-2 are two example neighbor raw frames: frame 2833 and frame 2834. b-3 The detected surface position curves of (b-1) and (b-2). b-4 The difference between the two curves in (b-3). b-5 The histogram of the counts of the difference value of (b-4). b-6 data points selected from (b-5) with the x-value (difference of the surface positions) of which fall within the range between −8 and 2. c The curve of the relative motion of every two neighbor frames. d The calculated motion curve, the tread of the motion curve, and the realigned motion curve. e The volume image after motion correction.
Extended field of view (200 µm × 200 µm × 3.2 mm) volumetric imaging (after motion correction and surface flattening) from the dorsal forearm of a 29-year-old male volunteer with the excitation wavelength of 880 nm
a 3D image of the motion-corrected volume. b Orthogonal view of the 3D image in the yz and xy plane. The dashed yellow lines indicate the sectioning position. The three dashed rectangular boxes from left to right correspond to the position of (c–e). c The zoomed-in orthogonal view of a sweat gland with only the TPF channel (left rectangular box in b). The arrows point to a spinal sweat duct. d The zoomed-in orthogonal view of a hair follicle (central rectangular box in b). e The zoomed-in orthogonal view of a subvolume (right rectangular box in b). f Vertical plane image of the volume with only TPF and SHG channels. g 3D image of the surface flattened volume. h Vertical plane image of the surface flattened volume. i–k Horizontal plane image of the surface flattened volume at different depths. Color is coded in green for RCM, red for TPF, and magenta for SHG. Color is coded in green for RCM, red for TPF, and magenta for SHG. Scale bar 100 µm.

+3

Motion-tolerant 3D volumetric multimodality microscopy imaging of human skin with subcellular resolution and extended field-of-view
  • Article
  • Full-text available

February 2025

·

23 Reads

Communications Biology

Zhenguo Wu

·

Yunxian Tian

·

Jianhua Zhao

·

[...]

·

The skin, the body’s largest organ, has a heterogeneous structure with various cell types and tissue layers. In vivo noninvasive 3D volumetric imaging with multi-contrast, high resolution, a large field-of-view (FOV), and no-motion artifacts is crucial for studying skin biology and diagnosing/evaluating diseases. Traditionally high-resolution in vivo skin microscopy methods capture images in the en-face (xy) plane parallel to the skin surface but are affected by involuntary motion, particularly during large-area volumetric data acquisition using xy-z mosaicking. In this work, we developed an xz-y imaging method that acquires images in the vertical (xz) plane and extends the FOV by moving the skin laterally along the y-direction. This approach is conceived based on our observation that involuntary skin movements are mostly along the vertical direction. Combined with a unique motion correction method, it enables 3D image reconstruction with subcellular resolution and an extended FOV close to a centimeter (8 mm). A multimodality microscopy system using this method provides simultaneous reflectance confocal, two-photon excited fluorescence, and second harmonic generation imaging, enabling multi-contrast capabilities. Using this system, we captured histology-like features of normal skin, vitiligo, and melanoma, demonstrating its potential for in vivo skin biology studies, clinical diagnosis, treatment planning and monitoring.

Download

The Impact of Skin Cancer: Evaluating Family Physician Referrals for Skin Cancer and Diagnostic Outcomes

November 2024

·

62 Reads

Objectives: The study aims to evaluate the impact of skin cancer on the healthcare system by evaluating the outcomes of family physician referrals for potential skin cancer cases through patients assessed at the Skin Care Centre (Vancouver General Hospital). The primary goal is to determine the proportion of patients diagnosed with skin cancer following these referrals. Methods: A retrospective chart review was conducted for patients referred to the Skin Care Centre between May 2021 and May 2022. Data were collected on patient demographics, reasons for referral, and final diagnoses. The analysis focused on quantifying the number of patients diagnosed with skin cancer. Results: Preliminary findings indicate that out of 304 patients referred for skin cancer concerns, only 44 were diagnosed with skin cancer. The accuracy of family physicians in diagnosing keratinocyte cancers was notably higher than for melanoma. Specifically, 83 patients were referred for a keratinocyte carcinoma, and 43 were confirmed with the diagnosis, whereas 174 patients were referred for melanoma and only one was diagnosed with melanoma. The study also revealed demographic trends, with older patients being more likely to receive a skin cancer diagnosis. Conclusion: The findings highlight that the costs on the healthcare system is higher when considering patients being assessed for skin cancer that end up having benign lesions. The costs of patients with benign skin lesions are usually omitted when calculating the total cost of skin cancer on the healthcare system. Further research will focus on comparing these findings with additional patients and study cohorts.


Abstract Title: The Impact of Skin Cancer: Evaluating Family Physician Referrals for Skin Cancer and Diagnostic Outcomes

November 2024

·

12 Reads

Objectives: The study aims to evaluate the impact of skin cancer on the healthcare system by evaluating the outcomes of family physician referrals for potential skin cancer cases through patients assessed at the Skin Care Centre (Vancouver General Hospital). The primary goal is to determine the proportion of patients diagnosed with skin cancer following these referrals. Methods: A retrospective chart review was conducted for patients referred to the Skin Care Centre between May 2021 and May 2022. Data were collected on patient demographics, reasons for referral, and final diagnoses. The analysis focused on quantifying the number of patients diagnosed with skin cancer. Results: Preliminary findings indicate that out of 304 patients referred for skin cancer concerns, only 44 were diagnosed with skin cancer. The accuracy of family physicians in diagnosing keratinocyte cancers was notably higher than for melanoma. Specifically, 83 patients were referred for a keratinocyte carcinoma, and 43 were confirmed with the diagnosis, whereas 174 patients were referred for melanoma and only one was diagnosed with melanoma. The study also revealed demographic trends, with older patients being more likely to receive a skin cancer diagnosis. Conclusion: The findings highlight that the costs on the healthcare system is higher when considering patients being assessed for skin cancer that end up having benign lesions. The costs of patients with benign skin lesions are usually omitted when calculating the total cost of skin cancer on the healthcare system. Further research will focus on comparing these findings with additional patients and study cohorts. Methods: A retrospective chart review was conducted for patients referred to the Skin Care Centre between May 2021 and May 2022. Data were collected on patient demographics, reasons for referral, and final diagnoses. The analysis focused on quantifying the number of patients diagnosed with skin cancer.


In vivo micro-Raman spectroscopy from an arbitrary-shaped region of interest under simultaneous reflectance confocal imaging guidance

November 2024

·

23 Reads

Confocal Raman spectroscopy (CRS) has been widely used for noninvasive biochemical analysis of biological tissue in vivo. Currently, reflectance confocal microscopy (RCM) has been integrated with CRS as visual guidance for a square-/rectangular-shaped region of interest or a point of interest micro-Raman measurement. However, biological structures often have various morphologies; it is highly desirable to acquire a representative Raman spectrum from and only from the area that covers the whole target biological micro-structure. To achieve this goal, we developed a method to acquire Raman spectrum from a region of interest with any (arbitrary) shape under simultaneous RCM imaging guidance.




Fig. 1. Example of biopsy-confirmed melanoma cases from the EDRA dataset, accompanied by attributes outlined in the 7-point checklist.
Fig. 2. Conditional probability matrix for melanoma and 7PCL attributes. Each cell in the matrix represents the probability of one attribute given the presence of another. The rows correspond to the conditions, and the columns represent the attributes conditioned upon.
Fig. 5. The plot illustrates the weighted connections in the training set of the EDRA dataset. while (a) focuses on a specific pair, ATP-VS and IR-PIG. Yellow lines represent connections from ATP-VS to the other nodes, while cyan lines represent connections from IR-PIG. Variations in line thickness highlight differences in directed weights between the nodes. Figure (b) displays the overall connectivity map among all eight nodes.
AI-Enhanced 7-Point Checklist for Melanoma Detection Using Clinical Knowledge Graphs and Data-Driven Quantification

July 2024

·

76 Reads

The 7-point checklist (7PCL) is widely used in dermoscopy to identify malignant melanoma lesions needing urgent medical attention. It assigns point values to seven attributes: major attributes are worth two points each, and minor ones are worth one point each. A total score of three or higher prompts further evaluation, often including a biopsy. However, a significant limitation of current methods is the uniform weighting of attributes, which leads to imprecision and neglects their interconnections. Previous deep learning studies have treated the prediction of each attribute with the same importance as predicting melanoma, which fails to recognize the clinical significance of the attributes for melanoma. To address these limitations, we introduce a novel diagnostic method that integrates two innovative elements: a Clinical Knowledge-Based Topological Graph (CKTG) and a Gradient Diagnostic Strategy with Data-Driven Weighting Standards (GD-DDW). The CKTG integrates 7PCL attributes with diagnostic information, revealing both internal and external associations. By employing adaptive receptive domains and weighted edges, we establish connections among melanoma's relevant features. Concurrently, GD-DDW emulates dermatologists' diagnostic processes, who first observe the visual characteristics associated with melanoma and then make predictions. Our model uses two imaging modalities for the same lesion, ensuring comprehensive feature acquisition. Our method shows outstanding performance in predicting malignant melanoma and its features, achieving an average AUC value of 85%. This was validated on the EDRA dataset, the largest publicly available dataset for the 7-point checklist algorithm. Specifically, the integrated weighting system can provide clinicians with valuable data-driven benchmarks for their evaluations.


Figure 1. Methodology for precise microregistration that enables quantitative tracking of skin cells in vivo. (a) Illustration of the challenge for relocalization when the microscope probe is reattached to the skin at successive imaging sessions; (b) example of surface marker design, which has serrated edges that serve as reference markers for microregistration; (c) reflectance confocal microscopy (RCM)
Precise Serial Microregistration Enables Quantitative Microscopy Imaging Tracking of Human Skin Cells In Vivo

July 2024

·

71 Reads

Cells

We developed an automated microregistration method that enables repeated in vivo skin microscopy imaging of the same tissue microlocation and specific cells over a long period of days and weeks with unprecedented precision. Applying this method in conjunction with an in vivo multimodality multiphoton microscope, the behavior of human skin cells such as cell proliferation, melanin upward migration, blood flow dynamics, and epidermal thickness adaptation can be recorded over time, facilitating quantitative cellular dynamics analysis. We demonstrated the usefulness of this method in a skin biology study by successfully monitoring skin cellular responses for a period of two weeks following an acute exposure to ultraviolet light.


Improving skin cancer detection by Raman spectroscopy using convolutional neural networks and data augmentation

June 2024

·

127 Reads

·

3 Citations

Background Our previous studies have demonstrated that Raman spectroscopy could be used for skin cancer detection with good sensitivity and specificity. The objective of this study is to determine if skin cancer detection can be further improved by combining deep neural networks and Raman spectroscopy. Patients and methods Raman spectra of 731 skin lesions were included in this study, containing 340 cancerous and precancerous lesions (melanoma, basal cell carcinoma, squamous cell carcinoma and actinic keratosis) and 391 benign lesions (melanocytic nevus and seborrheic keratosis). One-dimensional convolutional neural networks (1D-CNN) were developed for Raman spectral classification. The stratified samples were divided randomly into training (70%), validation (10%) and test set (20%), and were repeated 56 times using parallel computing. Different data augmentation strategies were implemented for the training dataset, including added random noise, spectral shift, spectral combination and artificially synthesized Raman spectra using one-dimensional generative adversarial networks (1D-GAN). The area under the receiver operating characteristic curve (ROC AUC) was used as a measure of the diagnostic performance. Conventional machine learning approaches, including partial least squares for discriminant analysis (PLS-DA), principal component and linear discriminant analysis (PC-LDA), support vector machine (SVM), and logistic regression (LR) were evaluated for comparison with the same data splitting scheme as the 1D-CNN. Results The ROC AUC of the test dataset based on the original training spectra were 0.886±0.022 (1D-CNN), 0.870±0.028 (PLS-DA), 0.875±0.033 (PC-LDA), 0.864±0.027 (SVM), and 0.525±0.045 (LR), which were improved to 0.909±0.021 (1D-CNN), 0.899±0.022 (PLS-DA), 0.895±0.022 (PC-LDA), 0.901±0.020 (SVM), and 0.897±0.021 (LR) respectively after augmentation of the training dataset (p<0.0001, Wilcoxon test). Paired analyses of 1D-CNN with conventional machine learning approaches showed that 1D-CNN had a 1–3% improvement (p<0.001, Wilcoxon test). Conclusions Data augmentation not only improved the performance of both deep neural networks and conventional machine learning techniques by 2–4%, but also improved the performance of the models on spectra with higher noise or spectral shifting. Convolutional neural networks slightly outperformed conventional machine learning approaches for skin cancer detection by Raman spectroscopy.


Single source CARS-based multimodal microscopy system for biological tissue imaging [Invited]

December 2023

·

30 Reads

·

4 Citations

A coherent anti-Stokes Raman scattering (CARS)-based multimodality microscopy system was developed using a single Ti:sapphire femtosecond laser source for biological imaging. It provides three complementary and co-registered imaging modalities: CARS, MPM (multiphoton microscopy), and RCM (reflectance confocal microscopy). The imaging speed is about 1 frame-per-second (fps) with a digital resolution of 1024 × 1024 pixels. This microscopy system can provide clear 2-dimensional and 3-dimensional images of ex-vivo biological tissue samples. Its spectral selection initiates vibrational excitation in lipid cells (approximately 2850 cm⁻¹) using two filters on the pump and Stokes beam paths. The excitation can be tuned over a wide spectral range with adjustable spectral filters. The imaging capability of this CARS-based multimodal microscopy system was demonstrated using porcine fat, murine skin, and murine liver tissue samples.


Citations (61)


... Deep neural networks require a large number of cases for training. A number of data enhancement strategies have been proposed for image analysis, such as flipping, color space [16], panning, rotation, noise injection [17], image blending, random cropping, and generative adversarial networks [18]. In this paper, noise injection is used to add a small amount of random noise to the input data that is small and consistent with the distribution of the original data, as opposed to the data collected by the data glove. ...

Reference:

A Static Sign Language Recognition Method Enhanced with Self-Attention Mechanisms
Improving skin cancer detection by Raman spectroscopy using convolutional neural networks and data augmentation

... PDT involves the application of a photosensitizer to the skin, which, when activated by light, leads to the production of ROS and the destruction of cancer cells. PDT is effective, particularly in the treatment of superficial cancers and precancerous lesions such as solar keratosis [35][36]. ...

Incidence and Profile of Skin Cancers in Patients Following Ultraviolet Phototherapy Without Psoralens- a retrospective cohort study
  • Citing Article
  • December 2023

Journal of the American Academy of Dermatology

... The invited review paper highlights technologies for depth scanning in miniature optical imaging systems, which is crucial for translational and intravital imaging for clinical and neuroscience research [1]. In multimodal imaging, an invited research paper, that has also been selected as editors' pick, shows a microscopy system combining coherent anti-Stokes Raman scattering (CARS), multiphoton microscopy (MPM), and reflectance confocal microscopy (RCM), using a single Ti:sapphire laser pumping a supercontinuum generation [2]. Contributed research papers demonstrate multimodal 2-photon and 3-photon nonlinear microscopy using a frequency-doubled femtosecond fiber laser [3] or the supercontinuum generation pumped by a femtosecond fiber laser [4]. ...

Single source CARS-based multimodal microscopy system for biological tissue imaging [Invited]

... Moreover, the extent of inflammatory infiltration is correlated with disease progression, with more infiltration indicating faster progression of the disease. In such cases, treatment options may include oral minipulse therapy [20]. ...

| 1 Ghia 16 | Boon Kee Goh 17 | Pearl Grimes 18 | Somesh Gupta 19
  • Citing Article
  • September 2023

Journal of the European Academy of Dermatology and Venereology

... The treatment group additionally underwent NB-UVB (Waldmann Company, Germany) irradiation starting 2 weeks postsurgery, at a frequency of twice per week. The initial dose was 300-400 mJ/cm 2 , with a 20% increment per treatment, for a total of 30 treatments [7]. ...

Worldwide expert recommendations for the diagnosis and management of vitiligo: Position statement from the international Vitiligo Task Force-Part 2: Specific treatment recommendations
  • Citing Article
  • September 2023

Journal of the European Academy of Dermatology and Venereology

... Increasing evidence suggests combination therapy with tazarotene cream and NBUVB phototherapy provides faster and more robust clinical clearing of psoriatic skin than either single treatment entity alone, although head-to-head clinical studies are warranted. Also, the use of prior or simultaneous UVB therapy with ustekinumab may result in faster clinical results [25][26][27]. Although whole body UV irradiation is potentially hazardous, stable plaque psoriasis is one disease where the benefits of NBUVB therapy outweigh the risks. ...

Topical, Ablative and Light-Based Therapies for Non-Melanoma Skin Neoplasms
  • Citing Chapter
  • January 2023

... Support Vector Machines (SVMs) & Deep Learning Models: SVMs [10][11][12] and Deep Learning Models that use CECNN have been extensively investigated in the detection of skin cancer. SVMs seek to identify an optimal hyperplane that distinguishes between distinct classes of skin lesions using extracted features. ...

Cutaneous Porphyrin Exhibits Anti-Stokes Fluorescence Emission Under Continuous Wave Laser Excitation

IEEE Journal of Selected Topics in Quantum Electronics

... These included vitamin D deficiency, psoriasis, type 1 and 2 diabetes mellitus, hypoand hyperthyroidism, ulcerative colitis, vitiligo, and systemic lupus erythematosus (SLE). Each condition was identified using validated ICD-9 and 10 coding, confirmation through primary care and hospital records, or through self-report during recorded interviews detailed in the UKBB (Supplemental Material) [11,[13][14][15][16][17][18]. ...

Validation of medical service insurance claims as a surrogate for ascertaining vitiligo cases

Archives of Dermatological Research

... In another study, Wang et al. discuss the importance of CBIR as an effective "diagnostic tool." Accordingly, authors utilized the pre-trained ResNet50 to extract the most effective features from the skin images [20]. In [21], Wang et al. fused deep features and metric learning for scene image recognition. ...

Multi-channel content based image retrieval method for skin diseases using similarity network fusion and deep community analysis
  • Citing Article
  • September 2022

Biomedical Signal Processing and Control

... A further limitation of the study is the subjective assessment of sun exposure, since this may imply recall bias. However, it should be noted that self-reported measures are widely used in epidemiological studies [90][91][92] and some studies have shown a good correlation between objective and subjective measures of sun exposure assessment 93,94 . ...

Assessment of sun‐safety education behavior via spectrophotometric evaluation: A preliminary study
  • Citing Article
  • December 2021

Photodermatology Photoimmunology and Photomedicine