Optical tomography with ultrasound localization for breast cancer diagnosis and treatment monitoring

Two-stage classification strategy for breast cancer diagnosis using ultrasound-guided diffuse optical tomography and deep learning

Significance

Ultrasound (US)-guided diffuse optical tomography (DOT) has demonstrated great potential for breast cancer diagnosis in which real-time or near real-time diagnosis with high accuracy is desired.

Aim

We aim to use US-guided DOT to achieve an automated, fast, and accurate classification of breast lesions.

Approach

We propose a two-stage classification strategy with deep learning. In the first stage, US images and histograms created from DOT perturbation measurements are combined to predict benign lesions. Then the non-benign suspicious lesions are passed through to the second stage, which combine US image features, DOT histogram features, and 3D DOT reconstructed images for final diagnosis.

Results

The first stage alone identified 73.0% of benign cases without image reconstruction. In distinguishing between benign and malignant breast lesions in patient data, the two-stage classification approach achieved an area under the receiver operating characteristic curve of 0.946, outperforming the diagnoses of all single-modality models and of a single-stage classification model that combines all US images, DOT histogram, and imaging features.

Conclusions

The proposed two-stage classification strategy achieves better classification accuracy than single-modality-only models and a single-stage classification model that combines all features. It can potentially distinguish breast cancers from benign lesions in near real-time.

Effect and correction of optode coupling errors in breast imaging using diffuse optical tomography

In diffuse optical tomography (DOT) and spectroscopy (DOS) using handheld probes, tissue curvature can cause bad fiber-to-tissue contact. Understanding and minimizing image artifacts caused by these coupling errors would significantly improve DOT and DOS image quality. In this work, we utilized Monte Carlo simulations and experiments with gelatin-Intralipid phantoms to systematically study the influence of source or detector (optode) coupling errors. Optode coupling errors can increase the amplitude and decrease the phase of the measured diffuse reflectance, creating artifacts in the reconstructed absorption maps, such as hot spots on the edges. We propose an outlier removal algorithm that can correct these image artifacts, and we demonstrate its performance using simulations, phantom experiments, and breast patient data acquired with bad probe contact due to a dense or small breast. Further, we designed and implemented a new resistance-type thin-film force sensor array that provides real-time optode coupling feedback and guides the outlier removal to minimize optode coupling errors. Our approaches and study results have significant implications for reducing image artifacts arising from handheld probes, which are commonly used with mobile and wearable DOT and DOS devices.

Hybrid time-domain and continuous-wave diffuse optical tomography instrument with concurrent, clinical magnetic resonance imaging for breast cancer imaging

Diffuse optical tomography has demonstrated significant potential for clinical utility in the diagnosis and prognosis of breast cancer, and its use in combination with other structural imaging modalities improves lesion localization and the quantification of functional tissue properties. Here, we introduce a hybrid diffuse optical imaging system that operates concurrently with magnetic resonance imaging (MRI) in the imaging suite, utilizing commercially available MR surface coils. The instrument acquires both continuous-wave and time-domain diffuse optical data in the parallel-plate geometry, permitting both absolute assignment of tissue optical properties and three-dimensional tomography; moreover, the instrument is designed to incorporate diffuse correlation spectroscopic measurements for probing tissue blood flow. The instrument is described in detail here. Image reconstructions of a tissue phantom are presented as an initial indicator of the system's ability to accurately reconstruct optical properties and the concrete benefits of the spatial constraints provided by concurrent MRI. Last, we briefly discuss how various data combinations that the instrument could facilitate, including tissue perfusion, can enable more comprehensive assessment of lesion physiology.

Three-Dimensional Highly Sensitive Diffusion Reflection-Based Imaging Method for the in Vivo Localization of Atherosclerosis Plaques Following Gold Nanorods Accumulation

In this work, we present a novel, simple, and highly accurate three-dimensional (3D) diffusion reflection (DR) imaging system and method for the detection of accumulation sites of gold nanorods (GNRs) within the tissue. GNRs are intensively used for diagnosis purposes of varied diseases, mainly because of their ability to well absorb visible light, which introduces them as terrific contrast agents in various imaging and theranostics methods. Lately, these GNRs unique absorption properties have served in DR intensity-based measurements, suggesting a novel diagnostic tool, DR-GNRs. In this paper, we show a new measurement system and method for DR, based on its radial collection from the tissue. These radial measurements enabled a unique 3D presentation of the DR-GNR, introducing the dimensions ρ for the radius, θ for the angle, and Γ for the reflected intensity. On the basis of the diffusion model, which enables to correlate between the sample's optical properties and its reflectance, a unique, radial map is presented. This map introduces the slopes of the DR curves in each measured angle, which are linearly correlated with the tissue's optical properties and with the GNRs concentrations within the tissue, thus enables the exact radial localization of the GNRs in the sample. We show the detection of macrophage accumulation in tissue-like phantoms, as well as the localization of unstable plaques in hyperlipidemic mice, in vivo. This highly accurate, powerful technology paves the way toward a real-time detection method that can be successfully integrated in the rapid increasing field of personalized medicine.

Note: Localization based on estimated source energy homogeneity

Acoustic signal localization is a complex problem with a wide range of industrial and academic applications. Herein, we propose a localization method based on energy attenuation and inverted source amplitude comparison (termed estimated source energy homogeneity, or ESEH). This inversion is tested on both synthetic (numerical) data using a Lamb wave propagation model and experimental 2D plate data (recorded with 4 accelerometers sensitive up to 26 kHz). We compare the performance of this technique with classic source localization algorithms: arrival time localization, time reversal localization, and localization based on energy amplitude. Our technique is highly versatile and out-performs the conventional techniques in terms of error minimization and cost (both computational and financial).

Frequency domain near-infrared multiwavelength imager design using high-speed, direct analog-to-digital conversion

Frequency domain near-infrared spectroscopy (FD-NIRS) has proven to be a reliable method for quantification of tissue absolute optical properties. We present a full-sampling direct analog-to-digital conversion FD-NIR imager. While we developed this instrument with a focus on high-speed optical breast tomographic imaging, the proposed design is suitable for a wide-range of biophotonic applications where fast, accurate quantification of absolute optical properties is needed. Simultaneous dual wavelength operation at 685 and 830 nm is achieved by concurrent 67.5 and 75 MHz frequency modulation of each laser source, respectively, followed by digitization using a high-speed (180  MS/s) 16-bit A/D converter and hybrid FPGA-assisted demodulation. The instrument supports 25 source locations and features 20 concurrently operating detectors. The noise floor of the instrument was measured at <1.4  pW/√Hz, and a dynamic range of 115+ dB, corresponding to nearly six orders of magnitude, has been demonstrated. Titration experiments consisting of 200 different absorption and scattering values were conducted to demonstrate accurate optical property quantification over the entire range of physiologically expected values.

Nanoparticle uptake by macrophages in vulnerable plaques for atherosclerosis diagnosis

The composition of atherosclerotic (AS) plaques is crucial concerning rupture, thrombosis and clinical events. Two plaque types are distinguished: stable and vulnerable plaques. Vulnerable plaques are rich in inflammatory cells, mostly only M1 macrophages, and are highly susceptible to rupture. These plaques represent a high risk particularly with the standard invasive diagnosis by coronary angiography. So far there are no non-invasive low-risk clinical approaches available to detect and distinguish AS plaque types in vivo. The perspective review introduces a whole work-flow for a novel approach for non-invasive detection and classification of AS plaques using the diffusion reflection method with gold nanoparticle loaded macrophages in combination with flow and image cytometric analysis for quality assurance. Classical biophotonic methods for AS diagnosis are summarized. Phenotyping of monocytes and macrophages are discussed for specific subset labelling by nanomaterials, as well as existing studies and first experimental proofs of concept for the novel approach are shown. In vitro and in vivo detection of NP loaded macrophages (MΦ). Different ways of MΦ labelling include (1) in vitro labelling in suspension (whole blood or buffy coat) or (2) labelling of short-term MΦ cultures with re-injection of MΦ-NP into the animal to detect migration of the cells in the plaques and (3) in vivo injection of NP into the organism.

Towards non-invasive characterization of breast cancer and cancer metabolism with diffuse optics

We review recent developments in diffuse optical imaging and monitoring of breast cancer, i.e. optical mammography. Optical mammography permits non-invasive, safe and frequent measurement of tissue hemodynamics oxygen metabolism and components (lipids, water, etc.), the development of new compound indices indicative of the risk and malignancy, and holds potential for frequent non-invasive longitudinal monitoring of therapy progression.

Diffuse Optical Monitoring of the Neoadjuvant Breast Cancer Therapy

Recent advances in the use of diffuse optical techniques for monitoring the hemodynamic, metabolic and physiological signatures of the neoadjuvant breast cancer therapy effectiveness is critically reviewed. An extensive discussion of the state-of-theart diffuse optical mammography is presented alongside a discussion of the current approaches to breast cancer therapies. Overall, the diffuse optics field is growing rapidly with a great deal of promise to fill an important niche in the current approaches to monitor, predict and personalize neoadjuvant breast cancer therapies.

Noninvasive diffuse optical monitoring of head and neck tumor blood flow and oxygenation during radiation delivery

This study explored using a novel diffuse correlation spectroscopy (DCS) flow-oximeter to noninvasively monitor blood flow and oxygenation changes in head and neck tumors during radiation delivery. A fiber-optic probe connected to the DCS flow-oximeter was placed on the surface of the radiologically/clinically involved cervical lymph node. The DCS flow-oximeter in the treatment room was remotely operated by a computer in the control room. From the early measurements, abnormal signals were observed when the optical device was placed in close proximity to the radiation beams. Through phantom tests, the artifacts were shown to be caused by scattered x rays and consequentially avoided by moving the optical device away from the x-ray beams. Eleven patients with head and neck tumors were continually measured once a week over a treatment period of seven weeks, although there were some missing data due to the patient related events. Large inter-patient variations in tumor hemodynamic responses were observed during radiation delivery. A significant increase in tumor blood flow was observed at the first week of treatment, which may be a physiologic response to hypoxia created by radiation oxygen consumption. Only small and insignificant changes were found in tumor blood oxygenation, suggesting that oxygen utilizations in tumors during the short period of fractional radiation deliveries were either minimal or balanced by other effects such as blood flow regulation. Further investigations in a large patient population are needed to correlate the individual hemodynamic responses with the clinical outcomes for determining the prognostic value of optical measurements.

Near-infrared diffuse correlation spectroscopy in cancer diagnosis and therapy monitoring

A novel near-infrared (NIR) diffuse correlation spectroscopy (DCS) for tumor blood flow measurement is introduced in this review paper. DCS measures speckle fluctuations of NIR diffuse light in tissue, which are sensitive to the motions of red blood cells. DCS offers several attractive new features for tumor blood flow measurement such as noninvasiveness, portability, high temporal resolution, and relatively large penetration depth. DCS technology has been utilized for continuous measurement of tumor blood flow before, during, and after cancer therapies. In those pilot investigations, DCS hemodynamic measurements add important new variables into the mix for differentiation of benign from malignant tumors and for prediction of treatment outcomes. It is envisaged that with more clinical applications in large patient populations, DCS might emerge as an important method of choice for bedside management of cancer therapy, and it will certainly provide important new information about cancer physiology that may be of use in diagnosis.

Transrectal ultrasound-integrated spectral optical tomography of hypoxic progression of a regressing tumor in a canine prostate

The objective of this study was to evaluate if transrectal optical tomography implemented at three wavelength bands for spectral detection could monitor changes of the hemoglobin oxygen saturation (StO2) in addition to those of the total hemoglobin concentration ([HbT]) in lesions of a canine prostate, including an induced tumor modeling canine prostate cancer. Near-infrared (NIR) optical tomography was integrated with ultrasound (US) for transrectal imaging. Multi-spectral detection at 705_nm, 785_nm and 808_nm rendered measurements of [HbT] and StO2. Canine transmissible venereal tumor (TVT) cells were injected into the right lobe of a dog's prostate gland, which had a pre-existing cyst in the left lobe. Longitudinal assessments of the prostate were performed weekly over a 63-day duration by NIR imaging concurrent with grey-scale and Doppler US. Ultrasonography revealed a bi-lobular tumor-mass regressing from day-49 to day-63. At day-49 this tumor-mass developed a hypoxic core that became larger and more intense by day-56 and expanded further by day-63. The tumor-mass presented a strong hyper-[HbT] feature on day-56 that was inconsistent with US-visualized blood flow. Histology confirmed two necrotic TVT foci within this tumor-mass. The cyst appeared to have a large anoxic-like interior that was greater in size than its ultrasonographically delineated lesion, and a weak lesional elevation of [HbT]. On day-56, the cyst presented a strong hyper-[HbT] feature consistent with US-resolved blood flow. Histology revealed acute and chronic hemorrhage in the periphery of the cyst. The NIR imaging features of two other TVT nodules and a metastatic lymph node were evaluated retrospectively. Transrectal US-integrated spectral optical tomography seems to enable longitudinal monitoring of intra-lesional oxygenation dynamics in addition to the hemoglobin content of lesions in the canine prostate.

Clustered targets imaged by optical tomography guided by ultrasound

Clustered small breast lesions may be present in the neighboring areas and are difficult to accurately resolve and quantify in diffuse optical tomography. In addition, larger cancers are often accompanied by clustered satellite lesions in the neighboring areas, which are also difficult to resolve and quantify. To improve the light quantification of clustered lesions, a new multi-zone reconstruction algorithm guided by co-registered ultrasound (US) was investigated using simulations, phantoms, and clinical examples. This method separated one larger region-of-interest (ROI) into several ROIs based on the location information provided by co-registered US. In general, the single-ROI method cannot resolve two smaller targets when their separations were less than 2.5 cm and the depth was greater than 2.0 cm. The multi-zone reconstruction method improved the resolving ability and reconstruction accuracy. As a result, two targets located at 2.5 cm depth with separation greater than 2.0 cm could be distinguished, and reconstruction improved by more than 20% as compared with that of the single-ROI method. When two targets, one larger and one smaller, were located closer to each other, the location of the reconstructed absorption mass was shifted toward the larger target and the quantification of the smaller target was limited.

Optical imaging in cancer research: basic principles, tumor detection, and therapeutic monitoring

Accurate and rapid detection of diseases is of great importance for assessing the molecular basis of pathogenesis, preventing the onset of complications, and implementing a tailored therapeutic regimen. The ability of optical imaging to transcend wide spatial imaging scales ranging from cells to organ systems has rejuvenated interest in using this technology for medical imaging. Moreover, optical imaging has at its disposal diverse contrast mechanisms for distinguishing normal from pathologic processes and tissues. To accommodate these signaling strategies, an array of imaging techniques has been developed. Importantly, light absorption, and emission methods, as well as hybrid optical imaging approaches are amenable to both small animal and human studies. Typically, complex methods are needed to extract quantitative data from deep tissues. This review focuses on the development of optical imaging platforms, image processing techniques, and molecular probes, as well as their applications in cancer diagnosis, staging, and monitoring therapeutic response.

Quantitative second harmonic generation imaging and modeling of the optical clearing mechanism in striated muscle and tendon

We have investigated the mechanisms and capabilities of optical clearing in conjunction with second harmonic generation (SHG) imaging in tendon and striated muscle. Our approach combines three-dimensional (3-D) SHG imaging of the axial attenuation and directional response with Monte Carlo simulation (based on measured bulk optical properties) of the creation intensity and propagation through the tissues. Through these experiments and simulations, we show that reduction of the primary filter following glycerol treatment dominates the axial attenuation response in both muscle and tendon. However, these disparate tissue types are shown to clear through different mechanisms of the glycerol-tissue interaction. In the acellular tendon, glycerol application reduces scattering by both index matching as well as increasing the interfibril separation. This results in an overall enhancement of the 3-D SHG intensity, where good agreement is found between experiment and simulation. Through analysis of the axial response as a function of glycerol concentration in striated muscle, we conclude that the mechanism in this tissue arises from matching of the refractive index of the cytoplasm of the muscle cells with that of the surrounding higher-index collagenous perimysium. We further show that the proportional decrease in the scattering coefficient mu(s) with increasing glycerol fraction can be well-approximated by Mie theory.