Diagnostic accuracy of tactile imaging in selecting patients with palpable breast abnormalities: a prospective comparative study

Abstract

Unnecessary referrals of patients with breast lumps represent a significant issue, since only a few patients actually have lumps when examined by a breast specialist. Tactile imaging (TI) is a novel modality in breast diagnostics armamentarium. The aim of this study was to assess TI's diagnostic performance and compare it to clinical breast examination (CBE). This is a prospective, blinded, comparative study of 276 consecutive patients. All patients underwent conventional imaging and tissue sampling if either a radiological or a palpable abnormality was present. Sensitivity, specificity and positive and negative predictive values for CBE and TI were calculated. Radiological findings and final diagnosis based on histology and/or cytology were used as reference standards. Receiver operator characteristic (ROC) curve analysis was also performed for each method. Sensitivity and specificity of TI in detecting radiologically proven abnormalities were 85.5 % and 35 %, respectively. CBE's sensitivity was 80.3 % and specificity 76 %. In detecting a histopathological entity according to histology/cytology, sensitivity was 88.2 % for TI and 81.6 % for CBE. Specificity was 38.5 % and 85.7 % for TI and CBE, respectively. These results suggest a trend towards higher sensitivity of TI compared to CBE but significantly lower specificity. Subgroup analysis revealed superior sensitivity of TI in detecting a histological entity in pre-menopausal women. However, CBE's overall performance was superior compared to TI's according to ROC curve analysis. Although further research is necessary, the use of TI by the primary care physician as a selection tool for referring patients to a breast specialist should be considered especially in pre-menopausal women.

Tactile, Orientation, and Optical Sensor Fusion for Tactile Breast Image Mosaicking

“…Further developments will allow data to be combined at the scan probe, making synchronization automatic and allowing the system to operate in real time. The further development of TI requires the following validations: The current TIS should be tested on 3D breast phantoms to form truly representative breast maps, where more complex material phantoms can be produced with inclusions [13]; The TIS should implement a suitable high resolution tactile array, with subsequent analysis of inclusions/lesions to verify that the TIS composite image is effective compared to a single static image for estimation of lesion properties using traditional TI methods [10]; and then the TIS should be clinically validated on patients to demonstrate the expected improvement in diagnostic/differential accuracy versus existing TI methods [1], [12] and other methods such as ultrasound. This will verify the effectiveness of the delivered metrics generally.…”

“…The correlation between frames was reliant on maintaining consistent scan pressures, as the tactile features visible in tactile images are stress dependent. This meant that tracking coherence was easily lost if the scan stress changed significantly, so this method was never used in clinical trials of breast TI [1], [12].…”

“…If a tactile image contains the boundaries of a suspicious tissue, meaning a suspected lesion is fully within a tactile image, then size, shape, and elasticity analysis can be performed on the lesion reasonably accurately within 10% and 20% respectively [10], important in cancer diagnosis and management [11]. This information about the lesion can be used to differentiate malignant and benign conditions [1], [12] which is estimated to reduce the benign biopsy rate by 23% with no missed cancers [1]. In current practice, TI is used as an optional screening tool to separate patients who may need further testing in a noninvasive and comfortable manner, and is generally more mature than other developing handheld tumor detection systems [13].…”