Health Data Analytics

Ongoing Projects

Validation of Volumetric Breast Density as a Risk Factor for Breast Cancer

Development of a prognostication model for breast cancer

Improving Lung Cancer Wait Times

Evaluating Cardiac Risk in Breast Cancer Patients using Artificial Intelligence


Validation of Volumetric Breast Density as a Risk Factor for Breast Cancer

This study aims to validate three-dimensional measures of breast density as a risk factor for breast cancer. We plan to examine if and how volumetric breast density (VBD) is associated with tumour characteristics, breast cancer subtypes and survival outcomes following treatment to better understand the biological mechanisms through which breast density influences breast tumour prognosis and to potentially provide a biomarker for risk of breast cancer recurrence.

Personalized mammography screening has emerged as the way of the future for improving breast cancer screening effectiveness. In order for individualized screening to be applicable in a clinical setting, one must first improve methods of evaluating individual risk by validating risk factors. Breast density, a measure of the dense tissue in the breast, has the potential to be used as a strong predictor for breast cancer risk. Strong evidence relating breast density and cancer risk has been established for 2-dimensional methods of estimating breast density, even with the subjective nature of that process. Considering that the breast is a 3-dimensional structure, efforts were made to develop methods to assess VBD. With the implementation of digital mammography in B.C., the quantitative analysis of mammograms to estimate VBD is now available. However, validation data to support the direct correlation of VBD to breast cancer risk for an ethnically diverse population like British Columbia is not available. 

Screening mammograms and risk assessment questionnaires will be collected from 190,000 women in BC over two years. Women who are asymptomatic and aged 40 and older will be eligible to participate in the study and will be invited at their routine bilateral, two-view, screening mammogram.

Screening mammograms and questionnaires will be collected from twenty-two digital screening sites. Women in the study will be followed-up for 10 years. VBD measurements will be determined for all mammograms collected using Quantra and Volpara. The study will then establish the relative breast cancer risk attributable to mammographic density. The study will then examine whether there is a relationship between breast density and tumour characteristics, breast cancer subtypes and survival outcomes following treatment. How well volumetric breast density represents the risk of developing breast cancer and predicts response to therapy and breast cancer recurrence, are critical pieces of information needed for a screening program to develop an effective personalized strategy to detect and treat breast cancer.

This study proposes to establish the direct link between VBD and breast cancer risk for women in British Columbia. Once substantiated, VBD, together with other risk factor information, will be necessary for the screening mammography program to develop effective personalized breast screening strategies such as ultrasound screening for women with dense breasts. Therefore the results of this study may have the potential to change the paradigm of early breast cancer detection from a uniform-screening method toward an individualized, risk/benefit based personal approach.


Development of a prognostication model for breast cancer

The goal of this study is to build a predictive model for risk of breast cancer development and prognostication that incorporates well-known risk factors (age, breast density, BMI, family history, etc.), other socioeceonomic and health risk factors, and mammographic features from previous mammograms.

Mammographic features have been shown to be a promising predictor to include when determining breast screening eligibility, as they have been linked to stratifications of breast cancer risk. For example, women with false-positive results in previous mammograms due to calcifications have been found to have an increased risk of breast cancer, as have women with mammographic features that change over time. It is hypothesized that incorporation of prior mammographic feature information, and other socioeceonomic and health risk factors extracted from CAIS into risk assessment may help clinicians develop effective strategies to further optimize personalized breast cancer screening.

Multivariate regression and classification techniques will be explored to find a suitable model to predict breast cancer risk of development and outcomes based on mammographic features. Using retrospective longitudinal analysis, the change in mammographic features over time will be examined in relation to breast cancer outcomes. It is expected that the addition of other socio-economic and health risk factors will enhance the prognostication model performance.

Early detection of cancer improves patient outcomes, as the disease is often much more treatable when found at an early stage. Therefore, improving screening guidelines through incorporation of risk factors is crucial to reducing breast cancer mortality. The model produced by this study may allow screening mammography programs to gain an understanding of the relationship between mammographic features and risk of breast cancer development and outcomes.


Improving Lung Cancer Wait Times

The primary objective of this study is to quantify and characterize wait times for surgical lung cancer patients from referral to treatment, focusing on the time between four major checkpoints.

  1. Date of referral to the thoracic group
  2. Date of initial consult with a thoracic surgeon
  3. Date patient is “ready to treat”
  4. Date of definitive surgical resection

The secondary objective is to evaluate these wait times by specific sub-populations, (ie. age, health service delivery area, stage of tumour) and determine if differences in access to care are apparent and, if they are, identify and implement ways to rectify these challenges. Access to care is also being examined through the calculation of surgical resection rates for different sub-populations, which can then be compared to national and international standards.

Lung cancer is the leading cause of cancer-related deaths worldwide and it is the most readily diagnosed cancer amongst Canadians contributing to up to one quarter of all cancer-related deaths. Patients diagnosed at an earlier lung cancer stage have better clinical outcomes. However, elongated wait times in the care pathway prior to and during diagnosis can allow time for progression of the tumor and may result in poorer patient prognosis. For this reason, investigating access to lung cancer care is of great interest. Ensuring equal and equitable access to lung cancer care is important not only because of the high prevalence of lung cancer, but also because of the time sensitivity of care and complex progression of treatment.

This proposed initiative uses existing methodology to accurately quantify wait times in the thoracic surgical lung cancer pathway. This will allow for the comparison of the fastest and slowest wait times, information that is needed to evaluate thoracic surgical care quality and determine areas of improvement. Currently, the presentation of wait time data is inconsistent between reports and the use of a quantile regression model for measuring these values can be used for quality assurance purposes. This evaluation will also identify and review ways to rectify any points of lag in the lung cancer treatment pathway. Furthermore, this initiative will enable quality assurance monitoring of diagnostic, treatment, and outcomes for surgical lung cancer patients.


Evaluating Cardiac Risk in Breast Cancer Patients using Artificial Intelligence

The purpose of this project is to use deep learning and natural language processing methodologies to create an automated cardiac risk assessment tool for use in evaluating cardiac risk in breast cancer patients.

There is a known link between cardiac conditions and breast cancer. Women with breast cancer, for instance, may have a higher risk of later developing cardiovascular conditions than those without breast cancer. Further, women diagnosed with breast cancer that have a prior history of cardiovascular disease may be at increased risk of mortality from cardiac causes. Initial findings (not yet published) from BC Cancer find similar results. This increase in cardiac-related mortality may be caused by radiation-induced cardiac toxicity, with increases in heart disease proportionate to radiation dosage to the heart, potentially lasting 20 years or more.

Breast cancer patients undergoing radiation therapy undergo a Computed Tomography (CT) scan of the chest for radiation dose calculation. These CT scans offer an opportunity to estimate coronary artery calcium (CAC) score, an independent predictor of coronary events which improves cardiovascular risk prediction in asymptomatic individuals. If a patient is found to have a significantly high CAC score, they may be referred for formal cardiovascular risk evaluation. However, manually performing calcium scoring on all chest CT scans of breast cancer patients is impractical due to the tedious and time consuming process. Recent developments in artificial intelligence methodologies, such as deep learning and natural language processing (NLP), provide an opportunity to automate cardiac risk assessments for breast cancer patients.

Deep neural networks are a type of representation machine learning model which employ multiple processing layers to automatically detect, identify, or classify features of interest present within a dataset. In recent years, deep neural network machine learning techniques have been successfully applied in numerous medical contexts, including diagnostics, radiology, and pathology. Radiology in particular has found many transformative uses for deep learning across various modalities not only in classification tasks, but also increasingly in segmentation tasks. Automatic semantic segmentation allows for specific objects— such as anatomic structures, tissues, or organs—to be identified, contoured, and labeled, such as in the automatic scoring of cardiac calcification of CT scans.

NLP is a type of machine learning that uses computational techniques to process unstructured free text data into structured data. Applications of this are cross-disciplinary, including uses such as language translation and social media and web mining. Many previous applications of NLP have been used in consumer contexts (e.g. Apple’s Siri, analysing web-based customer reviews). However, in recent years, there has been increased interest in using NLP across medical contexts, in particular for automated mining of electronic health records (EHR), which are laborious and time-consuming to analyse by hand. In the radiation oncology context, NLP can identify cancer cases, attributes, and outcomes, with real-world applications in surveillance and epidemiology.

We believe this work will pave the way in using artificial intelligence models to aid in cardiac risk predictions in breast cancer patients during routine clinical care. Future applications of this work could help prevent cardiac-related mortalities in breast cancer patients and survivors of breast cancer.