The aim of the study is to determine whether the use of the CEST sequence would have diagnostic performance equivalent to the reference method of T2* infusion with contrast injection in the diagnosis of radionecrosis of lung cancer brain metastases.

Brain metastases are frequent secondary sites in the evolution of cancers, with an incidence of all cancers combined of about 20% and up to 40% in the later phases of the disease. Among these metastases, the lung is the most common primary lesion (responsible for 20-56% of brain metastases). The overall prognosis for metastatic brain cancers is poor, with treatment-free survival of 1 to 2 months, and with treatment averaging 7-8 months. Their treatment is based on systemic medical treatment (chemotherapy, immunotherapy or targeted therapy depending on the status of the lesion) and on the other hand on the response to local treatment (surgery or radiotherapy). Radiotherapy is performed either as a first-line or post-operative depending on the resectability of the lesions and the operability of the patient.

Radionecrosis is a late complication (6 months to 2 years on average) and frequent (5-25%) of stereotactic radiotherapy. It may be promoted by the concomitant use of immunotherapy such as checkpoint inhibitors, which implies a probable increase in its incidence. Treatment of radionecrosis is based on corticosteroids; recent studies also propose Bevacizumab.

The distinction in MRI between tumor progression and radionecrosis is based on a multimodal approach. Indeed, conventional sequences alone have average performance (sensitivity 76% and specificity 59%). Several MRI methods have been evaluated, but their performance varies according to the studies. Cerebral perfusion is the most widely used method, requiring contrast injection with correct performance. However, there is no standardized perfusion value to directly extrapolate these results. MRI spectroscopy has also been studied, with correct performance but only evaluated on weak samples and retrospectively.

CEST is an MRI technique that uses endogenous contrast, i.e. does not require injection of contrast medium. It consists of specifically altering the signal of a molecular compound by causing saturation (i.e. cancellation of its signal) and studying the saturation of the solute (i.e. water) on contact. Indeed, due to a process called 'chemical saturation transfer', the signal from the water in contact with the targeted compound will also become partially saturated. The subtraction of the MRI signals acquired before and after saturation makes it possible to obtain a tissue mapping of the molecular compound initially targeted obtaining an indirect reflection of its concentration. Indeed, the macromolecular composition of tissues (inaccessible for physical reasons in spectroscopy) differs according to radionecrosis status or tumor progression, with more ""amide"" compounds in the second case.

This relatively recent development technique has been studied in primary and secondary brain tumors, especially in the context of radionecrosis, but mainly in primary brain tumors. It seems to allow a more precise and earlier detection of possible tumor progressions.

The diagnosis of radionecrosis is therefore a major step forward for the management of patients with irradiated brain metastases.

Inclusion Criteria:

• Patients > 18 years of age
• Histologically proven primary lung cancer
• Histologically proven or not brain metastases
• Irradiated metastases
• Inclusion in a treatment protocol for brain metastases by brain metastasis in toto or stereotactic or gamma-knife radiotherapy
• Morphological increase of one or more lesions of secondary brain metastases on a follow-up MRI
• Patients affiliated to a social security scheme

Exclusion Criteria:

• Opposition to the study
• Contraindication to MRI
• Refusal of imaging by the patient
• Patient with state medical aid (unless exemption from affiliation)
• Severe cognitive impairment making informed consent impossible
• Patients under guardianship or deprived of liberty