Different types of medical imaging response criteria: morphological measures

In oncology clinical trials, therapy is generally assessed using imaging response criteria involving imaging biomarkers. Imaging response criteria are used to define a “treatment response” status, used by the radiologist to evaluate whether the response to therapy is complete or partial or conversely, if the disease is not responding or even progressing.

Imaging biomarkers are characteristics extracted from images under study and may vary with tumor or cancer types and with imaging modalities. They can be qualitative or quantitative. Response criteria are created using specific imaging biomarkers with metrics and sets of rules, to provide accurate and reproducible measurements of tumors during therapy. For example, imaging response criteria may include objective measures like tumor size and a set of rules such as size thresholds above which tumors are considered target lesions.

The basis of RECIST1.1 criteria: unidimensional measurements

The most common way to assess tumor size is by defining a unidimensional measurement of the tumor’s longest diameter, also referred to as a linear size measurement. In clinical trials, patients’ CT or MR scans are analyzed by selecting lesions of appropriate size (also known as target lesions) and by measuring the longest diameter of each individual selected tumor. Unidimensional line measurement forms the basis of RECIST1.1 criteria, and therefore is the most used metric to assess efficacy in oncology clinical trials for solid tumors.

When using RECIST1.1, up to 5 target lesions can be selected at baseline, and the sum of their diameters (longest diameter for non-nodal lesions and short axis for nodal lesions) is calculated to reflect the tumor burden for the patient. After the patient undergoes treatment, the same target lesions are measured again, and the percentage of change in the sum of diameters, plus the qualitative assessment of non target lesions and potential presence of new lesions, are used to classify response into one of four following categories:

  • Complete Response (CR)
  • Partial Response (PR)
  • Stable Disease (SD)
  • Progressive Disease (PD)

Download our: Tumor assessment using RECIST 1.1 brochure here

Target lesions are selected by the radiologist and assisted by image analysis software, like Median Technologies’ iSee® lesion management solution, and can be used to objectively determine tumor size and accurately calculate the percentage of change in tumors among different timepoints.

Unidimensional measurement, avoiding areas of necrosis and focusing on viable tumor part, using contrast enhanced CT or MRI

Anatomic size measurements do not always appropriately capture a positive tumor response, particularly when the therapeutic agent under investigation stabilizes disease rather than causes tumor shrinkage, or when the study is performed in certain cancer types.  Specifically, clinical researchers found a poor association between clinical benefit and traditional methods for assessing therapeutic response in patients with hepatocellular carcinoma (HCC) who were treated with sorafenib (brand name Nexavar™, Bayer and Onyx Pharmaceuticals). [iii]  To take into account tumor necrosis and viable tumor response of HCC under treatment, a new HCC-specific metric was created that focused only on viable tumor tissue (mRECIST). Unidimensional measurements that avoid areas of necrosis and focus on viable tumor part have been shown to be more accurate indicators of treatment response in this specific patient population and is the size metric of choice for HCC using mRECIST criteria. [iii]

Bidimensional measurements, the metric of choice in brain tumors

For bidimensional measurements using CT or MRI, two lines are placed across an individual tumor. The first line lies across the longest diameter and the second lies across the longest perpendicular diameter. The two values are then multiplied together. This measurement is also referred to as a cross-sectional measurement. Bidimensional measurements are a component of the original WHO criteria, but they have since been largely replaced with the unidimensional measurements of RECIST for most solid tumors, as single line measurements are easier to perform and therefore subject to less variability. [ii] However, bidimensional measurements are still the metric of choice in brain tumors (glioma, astrocytoma) and lymphoma, and remain an integral part of irRC, MacDonald, RANO, and Cheson criteria.

Volumetric measurements and QIBA Profile

Tumor size and shape can also be assessed by 3-dimensional, volumetric measurements, which can be performed during routine CT or MR imaging using a segmentation software that calculates the area of a region over several slices. Change in tumor volume can be an earlier and more sensitive indicator of tumor response than simple linear measurements, particularly for irregularly shaped or morphologically complex tumors. [i] Although no formal response criteria have yet been established for volume measurements on CT images, a QIBA (Quantitative Imaging Biomarkers Alliance) Profile has been created to address issues of accuracy and reproducibility of this metric. The QIBA Profile states that a true volume change (with 95% confidence) has occurred when the measured change is >24% in tumors with a longest diameter of 50-100 mm, >29% in tumors with a longest diameter of 35-49 mm, and >39% in tumors with a longest diameter of 10-34 mm. [iv]

Measurement of the ROI/attenuation coefficient on CT

A tumor region of interest (ROI) is a manually or computer-assisted drawn line around the tumor mass on a CT image that separates tumor from the surrounding normal tissue. Within the ROI, the attenuation coefficient, which is based on CT-derived Hounsfield units (a quantity commonly used to express standardized CT numbers), can be calculated. Depending on the viable/necrotic part of this tumor various measures can be reported. This type of measure can then be considered both morphological (density) and functional (reflection of necrosis/viable part). Measurement of the ROI/attenuation coefficient is included as part of the Choi criteria for patients with gastrointestinal stromal tumors (GIST) and as part of the Size and Attenuation CT (SACT) or Morphology, Attenuation, Size, and Structure (MASS) criteria for patients with renal cell carcinoma (RCC).

The table below represents various morphological imaging criteria that are available for clinical trials in oncology.


Post authored in collaboration with Median Technologies’ Science and Medical Team

To know more about the fundamentals of medical imaging in oncology watch our Imaging 101 below


[i] Goldmacher GV, Conklin J. (2012) Brit J Clin Pharmacol. 73 : 846-854.

[ii] James K, Eisenhauer E, Christian M, et al. (1999) J Natl Cancer Inst. 91: 523-528.

[iii] Lencioni R, Llovet JM. (2010) Semin Liver Dis. 30: 52-60.

[iv] Quantitative Imaging Biomarkers Alliance. QIBA Profile: CT Tumor Volume Change for Advanced Disease. http://qibawiki.rsna.org/images/0/04/QIBA_CTVol_TumorVolumeChangeProfile_Consensus-20161121b.pdf.  Accessed May 2017.