Medical imaging is an essential tool in healthcare for diagnosis, treatment, and monitoring. Within an image, a lot of data is captured that can be used to assess the present condition of a patient and to evaluate change through time.
Medical imaging can be assessed qualitatively with a general assessment of shape and texture, and quantitatively for accurate and sensitive analysis. In this blog post, we will review the most frequent metrics used in quantitative imaging analysis.
Hounsfield Units (HU) are a universally used unit in Computed Tomography (CT). They are named after Sir Godfrey Hounsfield, winner of the 1979 Nobel Prize in medicine for the invention of the CT.
HU are a relative quantitative measurement of radiodensity, the property of a tissue to attenuate/absorb an x-ray beam. Distilled water has a value of 0 HU on the Hounsfield scale whereas air has a value of -1,000 HU. Upper limits include 1,000 for bones, and more than 3,000 for metals like steel or silver.
Table 1: Hounsfield Units
|
Air |
-1000 |
|
Fat |
-100 to -50 |
|
Water |
0 |
|
Muscle |
10 to 40 |
|
Blood |
40 |
|
Liver |
40 to 60 |
|
Bone |
1,000 |
With the Hounsfield scale, images are displayed in grey tones. Dense tissues have positive values and appear bright; less dense tissues absorb less x-rays, have negative values, and appear dark.
The Hounsfield scale is a relative scale and is dependent on several CT parameters such as the reconstructing algorithm or the x-ray kilovoltage. Beam hardening artifact also affects the determination of HU units. Nevertheless, HU is a useful metric in clinical practice to aid in the interpretation of images. It can be used in a wide range of indications, including but not limited to the evaluation of:
In addition, HU values can be used in oncology for the monitoring of tumor response. Indeed, the CHOI criteria for GIST proposes to include tumor attenuation in addition to size for the evaluation of tumor response [1].
The Standardized Uptake Value or the Standard Uptake Value (SUV) is a dimensionless metric used to evaluate tracer uptake on Positron Emission Tomography (PET) imaging. It helps physicians distinguish between “normal” and “abnormal” levels of uptake. It is normalized for body weight and injected dose. If the dose is uniformly distributed over the entire body, the SUV value everywhere in the body would be 1.0.
SUV =
|
Activity concentration in the tissue
|
| Injected activity /body size |
SUV is a relative quantitative measurement with many factors impacting the result, such as patient size, time of measurement, dose extravasation, partial volume effects, and reconstruction parameters. To address these challenges, the calculation of SUV can be adapted; for example, with the substitution of body weight by lean body mass.
SUVs are of major interest in clinical studies as they can be used to accurately assess tumor response. Standard oncology criteria are often based on tumor size. However, changes in size due to therapy can take months, causing a delay in therapeutic decisions. Change in SUV can occur much earlier, providing an alternative approach to tumor response assessment. PERCIST criteria, for example, assess metabolic response by evaluating SUL, the standardized uptake value corrected for lean body mass, at each time point. However, in the context of clinical trials, to ensure the comparability of results across patients and sites, it is crucial to standardize image acquisition protocols and include an equipment calibration step in the site qualification process.
Radiomics is a quantitative approach extracting all possible features from an image using analysis methods from the field of artificial intelligence (AI). AI is indeed needed to evaluate the high amount of information that can be found in an image. In oncology, radiomics is of interest because it can capture disease-specific features that are imperceptible to the human eye.
However, there are many limitations to radiomics, including lack of standardization and proper validation. Radiomics is also influenced by variations in patients and equipment. There is a great deal of research in this field that addresses these challenges and aims to include radiomic tools in the radiological workflow.
Keosys R&D team also uses AI to extract imaging metrics that can be used for tumor response assessments. PhD research scientist Noémie Moreau recently published her work on the automatic segmentation of metastatic breast cancer lesions on 18F-FDG PET/CT longitudinal acquisitions from which imaging biomarkers were extracted to determine tumor response including SULpeak. PhD research scientist Constance Fourcade focused on improving image registration to evaluate lesion change more precisely through time. From the method proposed, the image biomarkers extracted helped create a tool wherein tumor response is color-coded on the image displayed. This tool can help a physician assess tumor burden more quickly and more accurately.
Find out more about how we can use imaging metrics in your trials by contacting our sales team at sales@keosys.com
[1] H. Choi et al., “Correlation of Computed Tomography and Positron Emission Tomography in Patients with Metastatic Gastrointestinal Stromal Tumor Treated at a Single Institution with Imatinib Mesylate: Proposal of New Computed Tomography Response Criteria,” J. Clin. Oncol., vol. 25, no. 13, pp. 1753–1759, May 2007, doi: 10.1200/JCO.2006.07.3049.