Emerging breast imaging technologies on the horizon

Abstract

Early detection of breast cancers by mammography in conjunction with adjuvant therapy have contributed to reduction in breast cancer mortality. Mammography remains the ‘gold-standard’ for breast cancer screening, but is limited by tissue superposition. Digital breast tomosynthesis and more recently, dedicated breast computed tomography have been developed to alleviate the tissue superposition problem. However, all of these modalities rely upon x-ray attenuation contrast to provide anatomical images and there are ongoing efforts to develop and clinically translate alternative modalities. These emerging modalities could provide for new contrast mechanisms and may potentially improve lesion detection and diagnosis. In this article, several of these emerging modalities are discussed with a focus on technologies that have advanced to the stage of in vivo clinical evaluation.

1. Introduction

Mammography has been the ‘gold-standard’ for early detection and diagnosis of breast cancers. It has been intensely scrutinized, particularly in several clinical trials over the years for screening of asymptomatic women. The combination of screening mammography for early detection, and adjuvant therapy have contributed to reduction in breast cancer mortality¹. Independent statistical modeling by seven groups of investigators that formed the Cancer Intervention and Surveillance Modeling Network (CISNET) consortium showed that the combination of screening and adjuvant therapy reduced breast cancer mortality by approximately 30% (range: 25% to 38%), with mammography screening contributing to approximately 46% of this reduction (range: 28% to 65%)¹. The Swedish Organised Service Screening Evaluation group compared breast cancer mortality before and introduction of organized screening and reported an approximately 40% reduction among women screened^{2, 3}. The aforementioned studies were based on data that were collected before the introduction of digital mammography^4–8. The Digital Mammographic Imaging Screening Trial (DMIST) showed that the screen-film and digital mammography had similar diagnostic accuracy for breast cancer screening, with digital mammography being more accurate in women with radiographically dense breasts, women under 50 years of age, and in premenopausal or perimenopausal women⁹. In the United States and in many countries and regions, screen-film mammography has almost been completely replaced by digital mammography. Mammography screening has been highly beneficial; however, the two-dimensional (2D) planar images result in tissue superposition. Hence, digital breast tomosynthesis (DBT) ^10–15 and more recently dedicated breast computed tomography (BCT) ^16–28 have been developed in an effort to alleviate the tissue superposition. DBT generates limited tomographic imaging of the breast while BCT provides for fully three-dimensional (3D) imaging without physical compression of the breast. In all of the aforementioned modalities, differences in x-ray linear attenuation coefficients in breast tissue contribute to the image contrast and this enables anatomic imaging to identify potentially malignant lesions in the breast.

Among commonly used breast imaging modalities, ultrasound is routinely and most often used for diagnostic evaluation and for image guidance during tissue sampling (biopsy). Ultrasound imaging can also be of value in assessing axillary lymph nodes following a pathological diagnosis of malignancy²⁹. Handheld ultrasound³⁰ and automated whole breast ultrasound (ABUS) ^31–33 have been investigated as adjuncts to mammography screening for women with dense breasts. The American College of Radiology Imaging Network (ACRIN) 6666 trial showed that addition of screening ultrasound to mammography in women with at least heterogeneously dense breast in at least one quadrant resulted in higher cancer detection rate but with higher number of false-positives. Dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) is predominantly used for post-diagnosis patient management such as for extent-of-disease evaluation^34–36 and monitoring of neoadjuvant therapy response^37–39. Screening dynamic contrast-enhanced DCE-MRI⁴⁰ is also recommended as an adjunct to mammography for women with high lifetime risk of breast cancer, defined as those with lifetime risk greater than approximately 20–25%. Addition of screening MRI for women with dense breasts has also been shown to increase cancer detection but with an increase in false-positives⁴¹. In some institutions, radionuclide imaging, also referred to as molecular breast imaging and positron emission tomography/computed tomography (PET/CT) are also used, with PET/CT used almost solely in women with pathology-verified diagnosis of malignancy.

While all of the aforementioned modalities are in clinical practice, there are several emerging modalities that are in various stages of research, development and clinical evaluation that could provide for new contrast mechanisms and potentially improve lesion detection and diagnosis. In this article, several of these emerging modalities and technologies are described. A brief overview of these modalities will be provided in this article. Emerging modalities and technologies that have advanced to the stage of in vivo clinical evaluation are described first. Examples of such modalities include near-infrared optical imaging and x-ray phase contrast imaging.

2. Near-infrared diffuse optical imaging

Near-infrared diffuse optical imaging utilizes optical photons (light) in the wavelength range of 650 to 1000 nm, which is mostly in the near-infrared region of the electromagnetic spectrum. When multiple measurements at differing wavelengths within the aforementioned range are utilized then it is referred to as near-infrared spectroscopy (NIRS) or diffuse optical spectroscopic imaging (DOSI)⁴² and enables functional imaging of several centimeter thick breast tissue. It provides non-invasive measurement of tissue absorption and scattering properties which allow for the determination of oxygenated and deoxygenated hemoglobin, water, and lipid concentration in tissues without the use of exogenous contrast agents. Thus, DOSI provides for functional imaging of oxygen consumption through derivation of total hemoglobin concentration and oxygen saturation^{43, 44}, as well as composition in terms of water and lipid. Depending upon the data acquisition method and the associated technology, NIRS can be classified into time-domain⁴⁵, frequency-domain⁴⁶ and continuous wave⁴⁷ imaging. Time-domain NIRS relies on time-of-flight measurements through the tissue and employs ultra-short laser-generated light pulses of the order of a few picoseconds and requires detectors with very high temporal resolution of sub-nanoseconds⁴⁸. A comprehensive review of time-domain NIRS is provided by Torricelli et al⁴⁸. Given the high temporal resolution requirements, the advanced instrumentation required for time-domain NIRS presents challenges for its practical implementation. Frequency-Domain NIRS utilizes intensity-modulated sources operating at frequencies in the order of several MHz or higher followed by detection of the amplitude and phase-shift after transmitting through the tissue. Currently, there are some limitations with respect to available technology in terms of wavelength range and system gain, which pose a challenge for quantifying water and lipid content in tissues⁴⁹. Continuous wave NIRS uses a broad wavelength light source that is often amplitude-modulated to a few kHz and is combined with phase-locked detection techniques to measure attenuation changes. Continuous-wave NIRS does not provide for patient-specific scattering measurements⁴⁹ and hence a combination of continuous-wave and frequency-domain NIRS is used to provide attenuation and scattering measurements^{49, 50}. NIRS systems can also be classified as handheld probes or fixed geometry systems, which can be parallel-plate for imaging under moderate compression or in the form of a cup surrounding the uncompressed breast.

NIRS derived parameters have been correlated with MRI derived breast density⁵¹ and mammographic breast density⁵² and can potentially be useful for breast-density related risk assessment. NIRS has been used in numerous studies on monitoring neoadjuvant therapy treatment response ^53–59. The American College of Radiology Imaging Network (ACRIN) 6691 trial⁵⁹ evaluated whether quantitative parameters from a DOSI system with a handheld probe could be of value in assessing neoadjuvant treatment response. The study utilized a tissue optical index metric that combines oxygenated hemoglobin, water and lipid content. Measurements were performed in study participants at the tumor location and at the corresponding non-tumor location in the contralateral breast at baseline, 5 to 10 days after first cycle to determine early changes, mid treatment, and after completion of treatment and prior to surgery. Among the 60 subjects who participated in the study, 34 study participants had complete data and were evaluable. Stratifying the evaluable cohort based on oxygen saturation at baseline, the study observed that for the cohort with high oxygen saturation at baseline (76.9% or greater) the relative change in the ratio of tissue optical index between tumor and non-tumor region in the contralateral breast between baseline and mid-treatment showed an accuracy of 0.83 (95% CI: 0.63–1.0) in predicting pathologically complete response. However, for subjects with oxygen saturation at baseline less than 76.9%, the tissue optical index ratio did not correlate with pathologically complete response. Considering the study included tumors of several subtypes, the chemotherapy regimens varied among participants and the relatively small sample size the observed results are encouraging and further studies are needed.

One major limitation of NIRS is the limited spatial resolution which is 1 cm are more, depending on breast thickness, and hence is challenging for detecting and diagnosing small lesions. To overcome this limitation, multimodality systems combining NIRS with ultrasound^{60, 61}, MRI^{62, 63}, mammography⁶⁴ and digital breast tomosynthesis^65–68 have been developed. In these multimodality systems, the images provided by each modality can be simply fused to provide co-registered anatomic and functional images. However, this approach does not improve the spatial resolution of NIRS images. An alternative to this method is to extract the spatial information from anatomic imaging modalities^{69, 70} and incorporate these as priors during NIRS reconstruction for improved spatial resolution. In addition to the aforementioned clinical applications, NIRS has also been used for assessing surgical margins for intraoperative imaging either alone or in conjunction with fluorescent markers⁷¹. The aforementioned description of various NIRS systems, technologies and clinical applications is not intended to be comprehensive. Several topical reviews on the use of NIRS for breast cancer imaging have been published^72–78.

3. X-ray phase contrast imaging

X-ray phase contrast imaging utilizes the wave nature of x-rays. When x-rays traverse through tissue, it not only undergoes attenuation but also refraction. The changes in phase and intensity of the x-ray wave-front as it traverses a given tissue, is related to the refraction and the attenuation properties of that tissue and is represented by its complex refractive index. The real part of the refractive index is related to refraction that contributes to the phase shift and the imaginary part is related to the x-ray attenuation. The phase shift is related to the electron density (number of electrons in a unit volume), whereas the x-ray attenuation is related to the atomic number (number of protons in the nucleus)⁷⁹. At x-ray photon energies suitable for breast imaging, the angular deviation is extremely small and is not discernible with conventional x-ray imaging systems. However, several phase-sensitive x-ray imaging approaches, generally referred to as x-ray phase contrast imaging have been developed and are being actively investigated for potential clinical translation.

X-ray phase contrast imaging can be broadly classified into propagation-based imaging, analyzer-based imaging, interferometric techniques, and other non-interferometric techniques^80–82. Each method has specific requirements for instrumentation including the x-ray source, imaging geometry and optical elements (crystals, gratings, coded apertures or masks). In propagation-based imaging,^83–85 once the x-ray wave-front traverses the tissue and after propagating for a large distance the distorted wave-front undergoes interference resulting in amplitude variations. The recorded image has a similar appearance of a radiographic image but with boundaries between tissues showing higher contrast, thus providing edge-enhancement. Propagation-based phase contrast imaging typically uses synchrotron sources. Typically, the distances between the breast and the detector are large, which in one study⁸⁶ that investigated 49 subjects was 2 meters. A prototype phase-contrast mammography system using a conventional x-ray tube was developed and used in an initial evaluation with 38 subjects⁸⁷. Subsequently, a large study in Japan included 3835 phase-contrast mammography exams⁸⁸. Comparing phase-contrast mammography with screen-film mammography showed similar recall rates (5.53% vs. 5%) and cancer detection rates (0.29% vs. 0.25%) ⁸⁸. The phase contrast mammography system utilized computed radiography^89–92 (CR) imaging plates, which after readout needed further processing and resampling to finer pixels before printing on to a thermographic film using a laser printer. These additional processing requirements along with the transition from CR to flat-panel detectors for mammography limited adoption to clinical practice.

Analyzer-based imaging⁹³ utilizes a pair of silicon crystals, one acting as a monochromator to provide near monochromatic x-ray beam and located before the breast tissue or sample and another acting as an analyzer crystal that reflects x-rays at a particular angle (Bragg peak) and is located downstream of the sample. Sampling the reflected x-rays centered at the Bragg peak, absorption and refraction contrast can be obtained⁹³. Generally, this approach using the crystal-pair requires scanning of the object, which takes substantial time. In a study using this approach for imaging surgical specimens⁹⁴, the scan times ranged from 4 to 200 seconds. While this initial study used a synchrotron source, development of a system using a 1-kW x-ray tube has been reported⁹⁵. The imaging time with this technique can be reduced by using monochromator-analyzer arrays instead of the single pair of crystals and by using a higher powered x-ray tube ⁹⁵.

Interferometric techniques use either crystals or gratings, which are periodic patterns that are either linear or two-dimensional. Development of crystal-based x-ray interferometer⁹⁶ has been reported and the approach has largely been limited to laboratory settings for initial evaluation of specimens⁹⁷. This approach typically requires synchrotron source. More recently, there has been large interest in exploring Talbot-Lau interferometry as it can be performed with conventional x-ray tubes and detectors but with added high-precision gratings^98–104. One of the advantages of this technique is that it can provide for 3 unique contrast mechanisms from a single acquisition¹⁰⁵, the attenuation contrast that is similar to a traditional mammogram, the differential phase contrast image which provides the projection of edges or structural boundaries, and the dark-field image which provides small-angle x-ray scattering (SAXS) information^106–108. Several studies have reported on the imaging of surgical breast specimens using laboratory built systems^{109, 110}. Development of a clinical prototype system¹¹¹ and ongoing work on optimizing¹¹² the system have been reported. To our knowledge, in vivo breast imaging using this technique is yet to be reported. There are several topical reviews on x-ray phase contrast imaging that provide additional insights into these methods and techniques^113–115.

4. Photoacoustic imaging

Photoacoustic imaging also referred to optoacoustic imaging uses a pulsed laser source operating in the near-infrared region to deliver light pulses to the breast tissue, which upon interaction, excites the breast tissue and results in the generation of a pressure wave that is recorded by an ultrasound transducer^{116, 117}. Similar to NIRS, this approach can provide for the distribution of hemoglobin oxygenation status but at a much higher resolution than diffuse optical imaging approaches. Various photoacoustic imaging devices have been developed including a handheld probe¹¹⁸, parallel-plate geometry with single-side light delivery^{119, 120} or dual-side light delivery^{121, 122}, array of detectors arranged in the form of a ring¹²³, cylinder¹²⁴, and hemisphere^{125, 126}. Many of these studies^{118–121, 123, 124} report on their initial experience with less than 30 subjects in the respective studies. In one relatively larger study, 29/39 cancers were visualized using a system utilizing parallel-plate geometry with dual-side illumination¹²². Topical reviews addressing the developments in photoacoustic imaging have been published^{127, 128}.

5. Summary

Several interesting approaches are being investigated that may have a potential for translating to breast imaging including x-ray acoustic imaging^{129, 130} that is analogous to photoacoustic imaging but utilizes x-ray photons instead of near-infrared optical photons for excitation and x-ray fluorescence computed tomography^{131, 132} for imaging high-atomic number contrast agents such as those based on iodine and gold. In addition to these new modalities, there is active scientific interest in developing energy-resolved photon-counting x-ray detectors^133–137 that can provide for simultaneously acquired dual-energy or multi-energy images, potentially enabling better discrimination of microcalcifications and wider adoption of intravenously administered iodinated contrast agents for contrast-enhancement studies. Consistent with the nature of emerging modalities, large-scale clinical studies are lacking. Additionally, the technology development is often an iterative process and translation to clinical practice requires a longer time frame. Nevertheless, these approaches provide for new avenues to explore so that these innovations could potentially have substantial clinical impact.