Patients with focal epilepsy presenting without lesions often do not respond to medical treatment and are less likely to receive resective surgery. The objective was to improve lesion detection and conspicuousness using a dedicated epilepsy research MRI protocol at a specialist surgical centre. We studied 40 controls and 43 patients with epilepsy who were deemed to be MRI-negative by virtue of earlier non-specialist MRI assessment. Localization of seizure onset had been evaluated using seizure semiology and EEG. We re-evaluated these patient cases with our protocol using (i) diagnostic assessment of multiple MRI sequences by an experienced neuroradiologist and (ii) an automated quantitative voxel-based lesion detection technique where T1-weighted image-based feature maps were entered into separate t-tests between individual patients and controls (uncorrected p<0.01 and p<0.05 FWE-cluster-corrected). 27 patients remained MRI-negative after investigation by a neuroradiologist. 16 patients (37%) had at least one potentially epileptogenic lesion (including focal cortical dysplasia and hippocampal sclerosis), 11 had lesions concordant with the likely epileptogenic region identified by EEG. Both feature maps together detected all sites of dysplasia (100%). A dedicated epilepsy research protocol can increase identification of patients with lesions by up to 37%. Diagnostics can be complemented by sensitive automated statistical testing.