Peritumoral Radiomics for Treatment Response Prediction

You are reading version 1 of this direction, frozen 2026-10-04. The latest version is version 2.

Version 1 of 275 papers100 conceptsfrozen 2026-10-04

In traditional oncology, the tumor is the protagonist. Radiologists and computer vision models typically focus on the "Region of Interest" (ROI)—the visible mass of the lesion itself. However, a growing body of research suggests that the most critical biological signals are often found not within the tumor core, but in the "neighborhood" immediately surrounding it.

This research direction, often termed peritumoral radiomics, treats the tissue surrounding a tumor as a rich source of high-dimensional data. By extracting quantitative features from the tumor-host interface, researchers are building models that predict treatment response, immune activity, and metastatic potential with far greater accuracy than tumor-centric models alone.

The Tumor is Not an Island

The intuition behind peritumoral radiomics is rooted in the concept of the tumor microenvironment (TME). A tumor does not exist in isolation; it actively remodels its surroundings to facilitate growth, evade the immune system, and invade healthy tissue.

As Nathaniel Braman and colleagues demonstrated in Intratumoral and peritumoral radiomics for the pretreatment prediction of pathological complete response to neoadjuvant chemotherapy based on breast DCE-MRI, analyzing the 2.5–5mm radius around a breast tumor captures textural patterns of resistance that are invisible within the tumor itself. This "buffer zone" contains the front lines of the biological battle between the cancer and the patient's body.

Mapping the Immune Battlefield

One of the most powerful applications of this approach is predicting a patient's response to immunotherapy. Immunotherapy success depends on whether the body’s immune cells—like T-cells and macrophages—have successfully infiltrated the area.

Researchers like Weicai Huang and Yuming Jiang have shown that the peritumoral region is often more indicative of immune cell infiltration than the tumor core. In Noninvasive imaging of the tumor immune microenvironment correlates with response to immunotherapy in gastric cancer, they found that CT-based radiomics scores from the peripheral ring of the tumor could non-invasively map the local neutrophil-to-lymphocyte ratio. Similarly, Zepang Sun argues in Comprehensive assessment of immune context and immunotherapy response via noninvasive imaging in gastric cancer that the peritumoral region provides the "spatial dynamics of immune evasion" that are essential for identifying who will benefit from anti-PD-1 therapies.

Capturing Infiltration and Edema

In brain and liver cancers, the "edge" of the tumor is rarely a hard boundary. Instead, it is a zone of diffuse infiltration.

  • Brain Tumors: In glioblastoma, the surrounding swelling (edema) is not just fluid; it contains infiltrating tumor cells. Shouchao Wang and Niha Beig have demonstrated that the interface between the enhancing tumor and the edematous region contains the most significant prognostic markers for survival. As noted in Radiogenomic analysis of hypoxia pathway is predictive of overall survival in Glioblastoma, hypoxia-driven biological changes are most detectable at this specific transition zone.
  • Invasive Margins: For breast and head and neck cancers, the "edge" features are predictive of metastasis. Jiacheng Hao developed the FECT framework to specifically target these boundary features, mimicking how a pathologist looks for a tumor breaching its basement membrane.

Mechanisms of Feature Extraction

To capture these signals, researchers have moved beyond simple manual segmentation. Several distinct mechanisms have emerged:

  1. Dilation and Annular Rings: Models automatically expand the tumor boundary outward by fixed increments (e.g., 5mm, 10mm, or 20mm). Shaowei Wu used concentric annular rings in Pretreatment radiomic biomarker for immunotherapy responder prediction in stage IB–IV NSCLC to capture a gradient of biological signals moving away from the tumor.
  2. Fuzzy Masking: Rather than a hard binary boundary, Marko Grahovac proposed using weighted probability masks in Machine learning predictive performance evaluation of conventional and fuzzy radiomics in clinical cancer imaging cohorts. This accounts for the "gray areas" at the tumor edge where pixels may contain both cancerous and healthy tissue.
  3. Habitat Imaging: This involves dividing the tumor and its surroundings into distinct metabolic or textural "habitats" using unsupervised clustering. Haozhe Huang and Ruihe Lai have used this to capture the internal and external heterogeneity of the tumor environment simultaneously.

Beyond the Margin: Systemic Radiomics

The most recent evolution of this field looks even further afield. If the peritumoral region captures the local environment, "extra-tumoral" radiomics captures the systemic state of the patient.

In A Novel Radiomics Approach for Predicting TACE Outcomes in Hepatocellular Carcinoma Patients Using Deep Learning for Multi-organ Segmentation, Krzysztof Bartnik showed that features from the spleen and kidneys could be more predictive of liver cancer treatment success than the tumor itself, as they reflect systemic health and portal hypertension. Similarly, Yangdi Wang found that the "creeping fat" (visceral adipose tissue) surrounding bowel lesions in Crohn’s disease contains the key to predicting drug resistance.

By shifting the focus from the lesion to the landscape, peritumoral radiomics provides a non-invasive "virtual biopsy" of the complex interactions between a disease and its host.

Papers in this version

Showing 60 of 75 papers.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 78 papers
  2. Version 1 · 2026-10-04 · 75 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_peritumoral_feature_extraction_460af016,
  title = {peritumoral feature extraction},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/peritumoral-feature-extraction/rdr_460af016031842be91942599d22f4159},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_peritumoral_radiomics_for_treatment_resp_460af016_v1,
  title = {Peritumoral Radiomics for Treatment Response Prediction},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/peritumoral-radiomics-for-treatment-response-prediction/rdr_460af016031842be91942599d22f4159/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}