Marcello Chang

Marcello Chang

Auto-generated profile, based on publicly available information

Marcello Chang

Department of Radiology, Incubator for Medical Mixed Reality at Stanford (IMMERS) Stanford University Palo Alto California USA

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 April 2020 in Clinical Cancer Research
Verified
Journal Article
Closed access
PDF file
IXIanto Lin Xi
Ianto Lin Xi
YZYijun Zhao
Yijun Zhao
RWRobin Wang
Robin Wang
MCMarcello Chang
Marcello Chang
SPSubhanik Purkayastha
Subhanik Purkayastha
SSS. William Stavropoulos
S. William Stavropoulos
Purpose: With increasing incidence of renal mass, it is important to make a pretreatment differentiation between benign renal mass and malignant tumor. We aimed to develop a deep learning model that distinguishes benign renal tumors from renal cell carcinoma (RCC) by applying a residual convolutional neural network (ResNet) on routine MR imaging. Experimental Design: Preoperative MR images (T2-weighted and T1-postcontrast sequences) of 1,162 renal lesions definitely diagnosed on pathology or imaging in a multicenter cohort were divided into training, validation, and test sets (70:20:10 split). An ensemble model based on ResNet was built combining clinical variables and T1C and T2WI MR images using a bagging classifier to predict renal tumor pathology. Final model performance was compared with expert interpretation and the most optimized radiomics model. Results: Among the 1,162 renal lesions, 655 were malignant and 507 were benign. Compared with a baseline zero rule algorithm, the ensemble deep learning model had a statistically significant higher test accuracy (0.70 vs. 0.56, P = 0.004). Compared with all experts averaged, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.60, P = 0.053), sensitivity (0.92 vs. 0.80, P = 0.017), and specificity (0.41 vs. 0.35, P = 0.450). Compared with the radiomics model, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.62, P = 0.081), sensitivity (0.92 vs. 0.79, P = 0.012), and specificity (0.41 vs. 0.39, P = 0.770). Conclusions: Deep learning can noninvasively distinguish benign renal tumors from RCC using conventional MR imaging in a multi-institutional dataset with good accuracy, sensitivity, and specificity comparable with experts and radiomics.

18 March 2015 in Glia
Verified
Journal Article
Closed access
PDF file
FGFuzheng Guo
Fuzheng Guo
JLJordan Lang
Jordan Lang
JSJiho Sohn
Jiho Sohn
EHElizabeth Hammond
Elizabeth Hammond
MCMarcello Chang
Marcello Chang
DPDavid Pleasure
David Pleasure

9 January 2019 in PLOS Genetics
Verified
Journal Article
Open access
PDF file
OSOrsolya Symmons
Orsolya Symmons
MCMarcello Chang
Marcello Chang
IMIan A Mellis
Ian A Mellis
JKJennifer M. Kalish
Jennifer M. Kalish
JPJihwan Park
Jihwan Park
ARArjun Raj
Arjun Raj
Extensive cell-to-cell variation exists even among putatively identical cells, and there is great interest in understanding how the properties of transcription relate to this heterogeneity. Differential expression from the two gene copies in diploid cells could potentially contribute, yet our ability to measure from which gene copy individual RNAs originated remains limited, particularly in the context of tissues. Here, we demonstrate quantitative, single molecule allele-specific RNA FISH adapted for use on tissue sections, allowing us to determine the chromosome of origin of individual RNA molecules in formaldehyde-fixed tissues. We used this method to visualize the allele-specific expression of Xist and multiple autosomal genes in mouse kidney. By combining these data with mathematical modeling, we evaluated models for allele-specific heterogeneity, in particular demonstrating that apparent expression from only one of the alleles in single cells can arise as a consequence of low-level mRNA abundance and transcriptional bursting. In mammals, most cells of the body contain two genetic datasets: one from the mother and one from the father, and—in theory—these two sets of information could contribute equally to produce the molecules in a given cell. In practice, however, this is not always the case, which can have dramatic implications for many traits, including visible features (such as fur color) and even disease outcomes. However, it remains difficult to measure the parental origin of individual molecules in a given cell and thus to assess what processes lead to an imbalance of the maternal and paternal contribution. We adapted a microscopy technique—called allele-specific single molecule RNA FISH—that uses a combination of fluorescent tags to specifically label one type of molecule, RNA, depending on its origin, for use in mouse kidney sections. Focusing on RNAs that were previously reported to show imbalance, we performed measurements and combined these with mathematical modeling to quantify imbalance in tissues and explain how these can arise. We found that we could recapitulate the observed imbalances using models that only take into account the random processes that produce RNA, without the need to invoke special regulatory mechanisms to create unequal contributions.

19 March 2023 in Otolaryngology–Head and Neck Surgery
Verified
Journal Article
Open access
FNFabian N. Necker
Fabian N. Necker
MCMarcello Chang
Marcello Chang
CLChristoph Leuze
Christoph Leuze
MTMichael C. Topf
Michael C. Topf
BDBruce L. Daniel
Bruce L. Daniel
FBFred M. Baik
Fred M. Baik
Head and neck surgeons often have difficulty in relocating sites of positive margins due to the complex 3‐dimensional (3D) anatomy of the head and neck. We introduce a new technique where resection specimens are 3D scanned with a smartphone, annotated in computer‐assisted design software, and immediately visualized on augmented reality (AR) glasses. The 3D virtual specimen can be accurately superimposed onto surgical sites for orientation and sizing applications. During an operative workshop, a surgeon using AR glasses projected virtual, annotated specimen models back into the resection bed onto a cadaver within approximately 10 minutes. Colored annotations can correspond with pathologic annotations and guide the orientation of the virtual 3D specimen. The model was also overlayed onto a flap harvest site to aid in reconstructive planning. We present a new technique allowing interactive, sterile inspection of tissue specimens in AR that could facilitate communication among surgeons and pathologists and assist with reconstructive surgery.