The poor prognosis in the early-stage of oesophageal carcinoma cancer is due to potential worsening of the disease (local relapse, metastasis), to insufficient efficacy and toxicity of actual treatments.

FDG-PET is a medical imaging modality allowing the quantification of the tumour glucose consumption. Then, this exam is used for pathology staging, target volume definition for RT, and treatment efficiency few months after CRT. Our assumption is that an FDG-PET exam during the course of CRT might be predictive of the treatment efficiency few months later.

In this study, we propose to perform 4 FDG-PET: first "PET1" before chemo-radiotherapy, second "PET2" during the chemo-radiotherapy (see RTEP1), third and fourth "PET3" "PET4" 3month and 12 month after the end of radiotherapy.

We will investigate the performances of FDG-PET performed during CRT for the prediction of the one-year patient heath outcome. If the predictive value of TEP2 is confirmed, we would be able to optimize the planning treatment during the course of the therapy.

Inclusion Criteria:

• Carcinoma épidermoïde of the oesophagus histologically proved, UICC Stage II B, III or IVa (TNM on 2002)
• Decision of treatment by radio concomitant chemotherapy (radiotherapy in classic spreading of 50Gy associated with a platinum chemotherapy.

Exclusion Criteria:

• Presence of a second evolutionary cancer in the previous three years
• index of performance OMS > 2
• Patients badly balanced diabetics

Diagnosed patients a carcinoma of the oesophagus histologiquement proved