Device and method for percutaneous breast lump diagnosis and intraoperative tumor positivity detection

Abstract

The present disclosure relates to a system (100) for diagnosing tissue specimens of the breast region of a subject, the system (100) includes a photoacoustic probe (102) having a cannula (112) to enclose an optical fiber cable (114) to direct illumination onto the tissue specimen. A quartz element (120) embedded at the tip of the cannula (112) restricts the flow of body fluid flowing through the cannula. A detector (118) coupled to the quartz element at the dorsal end detects the set of photoacoustic signals induced in the tissue specimen upon light excitation. A processor is configured to receive the set of photoacoustic signals, analyse the set of photoacoustic signals to extract a set of features and classify the extracted set of features as a normal, benign, pre-malignant and malignant state of the breast region.

Description

TECHNICAL FIELD

[0001] The present disclosure relates, in general, to cancer treatment, andmore specifically, relates to a device and method for percutaneous breast lumpdiagnosis and intraoperative tumor positivity detection by machine learningassistedphotoacoustic spectroscopy.BACKGROUND

[0002] Breast-conserving surgery (BCS) is an effective treatment for earlystagemalignancies as long as the resected tissue's margins are disease-free,according to consensus guidelines for patient management. However, its link toreoperation rates of 20-40% after inadequate tumor removal necessitates thedevelopment of an intraoperative surgical margin evaluation tool that providescellular, structural, and molecular information of the entire specimen surface to aclinically relevant depth. Currently, intraoperative techniques for detecting thebreast cancer margin are gold standards, histology, frozen sections, positronemission tomography (PET) scan and the like, which are time-consuming, andothers have lower specificity and sensitivity. This would result in a 20% to 40%reoperation rate following incomplete tumor removal.

[0003] An example of existing technology in the field of cancer treatment isrecited in literature, entitled "an innovative approach to reducing the incidence ofpositive margins found after lumpectomy". The literature describes the goal oflumpectomy surgery for breast cancer as to completely remove the tumour and haveclear margins, reducing the rates of local recurrence. The working principle dependson local electrical properties measurements. However, the existing technologymentioned above suffers from limitations that include low sensitivity, lowspecificity, and high false-positive rates. Another example is recited in literature,entitled "Intra-operative assessment of excised breast tumour margins usingClearEdge imaging device". The literature aims to remove breast cancer completelyand obtain clear margins, however, suffers from a lack of sensitivity and specificity.

[0004] Yet another example is recited in literature, entitled "Real-time,intraoperative detection of residual breast cancer in lumpectomy cavity walls usinga novel cathepsin-activated fluorescent imaging system". The literature obtainstumour-free surgical margins to prevent recurrence in breast-conserving surgery.However, the existing technology mentioned above suffers from limitations thatinclude the use of external chemical agents / fluorophores.

[0005] Therefore, it is desired to overcome the drawbacks, shortcomings, andlimitations associated with existing solutions, and develop a device that facilitatesreal-time breast lump diagnosis and intraoperative tumor positivity detection bymachine learning-assisted photoacoustic spectroscopy.OBJECTS OF THE PRESENT DISCLOSURE

[0006] An object of the present disclosure relates, in general, to cancertreatment, and more specifically, relates to a device and method for percutaneousbreast lump diagnosis and intraoperative tumor positivity detection by machinelearning assisted photoacoustic spectroscope.

[0007] Another object of the present disclosure is to provide a device with asingle-use, disposable medical-grade needle for percutaneous insertion into a breastlump diagnosis and detection of tumor positivity on the surface of the residualcavity after intraoperative tumor excision.

[0008] Another object of the present disclosure is to provide a device that mayprovide a real-time tissue diagnosis and detection of margin positivity.

[0009] Another object of the present disclosure is to provide a device with theintegration of machine learning that avoids the essential requirement of a trainedcytopathologist for the evaluation of Fine-Needle Aspiration Cytology (FNAC)slides, enabling surgeons to diagnose breast pathology independently in clinics inless time.

[0010] Another object of the present disclosure is to provide a device thatupdates machine-learning algorithms upon the arrival of new breast tumour cases.

[0011] Another object of the present disclosure is to provide a device thatprovides cloud-based training.

[0012] Another object of the present disclosure is to provide a device thatavoids the requirement of external agents / chemicals for detection.

[0013] Another object of the present disclosure is to provide a device thatincreases the specificity and sensitivity of the solution.

[0014] Yet another object of the present disclosure is to provide a device thatprevents unwanted surgery for benign lumps alleviating the anxiety of the patientswaiting for the FNAC results.SUMMARY

[0015] The present disclosure relates in general, to cancer treatment, andmore specifically, relates to a device and method for percutaneous breast lumpdiagnosis and intraoperative tumor positivity detection by machine learningassistedphotoacoustic spectroscopy. The main objective of the present disclosureis to overcome the drawback, limitations, and shortcomings of the existing systemand solution, by providing photoacoustic spectroscopy integrated machine learningto diagnose breast lumps, which is cell and tissue-specific and can be used to detectbreast tumor margin with high sensitivity. The proposed photoacoustic probe wouldeliminate the need for patients to be re-examined in the event of a false negativediagnosis.

[0016] The present disclosure relates to a system for diagnosing tissuespecimens of the breast region of a subject, the system includes a photoacousticprobe having a cannula to enclose an optical fiber cable. The optical fiber cable isadapted to direct illumination from an excitation source onto the tissue specimen ofthe breast region and to collect a set of photoacoustic signals therefrom. The opticalfiber cable emerging from the excitation source is passed into the cannula to directillumination into the tissue specimen of the breast region. The set of photoacousticsignals is collected through the photoacoustic probe and recorded in anoscilloscope, where the oscilloscope is coupled to the excitation source. A quartzelement embedded at the tip of the cannula restricts the flow of body fluid flowingthrough the cannula. A detector coupled to the quartz element at the dorsal end anddetects the set of photoacoustic signals induced in the tissue specimen upon lightexcitation. The set of photoacoustic signals recorded in the oscilloscope.

[0017] The processor operatively coupled with a memory, the memorystoring instructions executable by the processor to receive the set of photoacousticsignals from the oscilloscope. The processor pre-processes and analyses the set ofphotoacoustic signals to extract a set of features, the set of features pertaining toany or a combination of fibroadenoma and phyllodes tissue. The set ofphotoacoustic signals from fibroadenoma and phyllodes tissues are subjected to afeature selection algorithm to eliminate unnecessary information and retain usefulinformation. The feature selection algorithm uses a mutual information-basedempirical approach to select the features followed by feature importance rankingvalue.

[0018] Further, the processor can classify the extracted set of features basedon matching the extracted set of features with a reference set of features. Theclassification pertains to a set of tissue types of the breast region, the set of tissuetypes pertaining to any or a combination of benign, pre-malignant and malignantstates of the breast region. Based on a combination of classification of the extractedset of features, the processor is configured to determine a diagnosis for the breastregion of the subject. Thus, the present disclosure provides real-time tissuediagnosis in comparison to obtaining a sample and processing it in a histopathologylaboratory and prevents unwanted surgery for benign lumps alleviating the anxietyof the patients waiting for the FNAC results.

[0019] In addition, the processor is operatively coupled to a learning engine,the learning engine adapted to receive the pre-processed set of features and trainedto evaluate the set of tissue types as the benign, pre-malignant and malignant stateof the breast region. The learning engine is trained using historical data ofcorrelation of the received set of features of the tissue specimen. The device withthe integration of machine learning avoids the essential requirement of a trainedcytopathologist for the evaluation of FNAC slides, enabling surgeons to diagnosebreast pathology independently in clinics in less time. The proposed deviceincorporates photoacoustic spectroscopy and machine learning to detect malignantresidual tissues in the operating cavity and permit real-time confirmation of marginnegative status in the residual cavity after Wide Local Excision (WLE). The devicepermits the assessment of tumor on the surface / margins of the resected WLEspecimen.

[0020] Besides, the processor is operatively coupled to a cloud-based serverthat is configured to update the database periodically to maintain current cases anddata of the tissue specimen. Thus, the device updates machine-learning algorithmsupon the arrival of new breast tumour cases. Those skilled in the art wouldappreciate that as the conventional diagnosis method is avoided in the presentinvention, the repeat visit for the patient and the essential requirement of a trainedcytopathologist for the evaluation of FNAC slides can be avoided, enabling thesurgeon to diagnose breast pathology independently in clinics in less time. Further,the device avoids the requirement of external agents / chemicals for detection andincreases the specificity and sensitivity of the solution.

[0021] Various objects, features, aspects, and advantages of the inventivesubject matter will become more apparent from the following detailed descriptionof preferred embodiments, along with the accompanying drawing figures in whichlike numerals represent like components.BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The following drawings form part of the present specification and areincluded to further illustrate aspects of the present disclosure. The disclosure maybe better understood by reference to the drawings in combination with the detaileddescription of the specific embodiments presented herein.

[0023] FIG. 1A illustrates an exemplary system for diagnosing breast lumpsin real-time using the photoacoustic probe, in accordance with an embodiment ofthe present disclosure.

[0024] FIG. 1B illustrates an exemplary photoacoustic probe design for breastlump diagnosis, in accordance with an embodiment of the present disclosure.

[0025] FIG. 1C illustrates a schematic view of intra-operative margindetection using the photoacoustic probe, in accordance with an embodiment of thepresent disclosure.

[0026] FIG. 2 illustrates exemplary functional components of the proposedsystem in accordance with an embodiment of the present disclosure.

[0027] FIG. 3 illustrates an instrumentation setup used for obtainingphotoacoustic spectra from tissues at 281 nm pulsed laser excitations, in accordancewith an embodiment of the present disclosure.

[0028] FIG. 4A and 4B illustrate graphical view representing the raw timedomainphotoacoustic spectra belonging to Phyllodes and Fibroadenoma, inaccordance with an embodiment of the present disclosure.

[0029] FIG. 5A and 5B illustrate graphical view representing the typical pre10processed photoacoustic spectra belonging to Phyllodes and Fibroadenoma, inaccordance with an embodiment of the present disclosure.

[0030] FIG. 6 illustrates a graphical view representing the wavelet coefficientversus prediction rank values, in accordance with an embodiment of the presentdisclosure.

[0031] FIG. 7 illustrates a schematic view of the confusion matrix showingthe performance of the Support vector machine learning model, in accordance withan embodiment of the present disclosure.

[0032] FIG. 8 illustrates a flow chart of a method for diagnosing breast lumpsin real-time using the photoacoustic probe, in accordance with an embodiment ofthe present disclosure.DETAILED DESCRIPTION

[0033] The following is a detailed description of embodiments of thedisclosure depicted in the accompanying drawings. The embodiments are in suchdetail as to clearly communicate the disclosure. If the specification states acomponent or feature "may", "can", "could", or "might" be included or have acharacteristic, that particular component or feature is not required to be included orhave the characteristic.

[0034] As used in the description herein and throughout the claims thatfollow, the meaning of "a," "an," and "the" includes plural reference unless thecontext clearly dictates otherwise. Also, as used in the description herein, themeaning of "in" includes "in" and "on" unless the context clearly dictatesotherwise.

[0035] The present disclosure relates, in general, to cancer treatment, andmore specifically, relates to a device and method for percutaneous breast lumpdiagnosis and intraoperative tumor positivity detection by machine learningassistedphotoacoustic spectroscopy. The proposed system disclosed in the presentdisclosure overcomes the drawbacks, shortcomings, and limitations associated withthe conventional system by providing photoacoustic spectroscopy integrated withmachine learning to diagnose breast lumps, which is cell and tissue-specific and canbe used to detect breast tumor margin with high sensitivity. The proposedphotoacoustic probe would eliminate the need for patients to be re-examined in theevent of a false negative diagnosis. The present disclosure can be described inenabling detail in the following examples, which may represent more than oneembodiment of the present disclosure.

[0036] The advantages achieved by the device of the present disclosure canbe clear from the embodiments provided herein. The present disclosure relates tothe photoacoustic probe with a single-use, disposable medical grade needle forpercutaneous insertion into the breast lump diagnosis and detection of tumorpositivity on the surface of the residual cavity after intraoperative tumor excision toassess the status of negative margins in assessing breast cancer in vivo. The devicemay provide a real-time tissue diagnosis and detection of margin positivity incomparison to the conventional method of obtaining a sample and processing it ina histopathology laboratory, which usually takes 4 to 5 days. The integration ofmachine learning may avoid the essential requirement of a trained cytopathologistfor the evaluation of Fine-Needle Aspiration Cytology (FNAC) slides, enabling thesurgeon to diagnose breast pathology independently in clinics without the processesinvolved in obtaining and interpreting the FNAC. Further, the point of care devicemay facilitate repeat assessments when clinically indicated, during follow-up visits.Additionally, it can prevent unwanted surgery for benign lumps alleviating theanxiety of the patients waiting for the FNAC results. The description of terms andfeatures related to the present disclosure shall be clear from the embodiments thatare illustrated and described; however, the invention is not limited to theseembodiments only. Numerous modifications, changes, variations, substitutions, andequivalents of the embodiments are possible within the scope of the presentdisclosure. Additionally, the invention can include other embodiments that arewithin the scope of the claims but are not described in detail with respect to thefollowing description.

[0037] FIG. 1A illustrates an exemplary system for diagnosing breast lumpsin real-time using the photoacoustic probe, in accordance with an embodiment ofthe present disclosure.

[0038] Referring to FIG. 1A, the system 100 can include a photoacousticprobe 102 that is fabricated to diagnose the breast region e.g., breast lumps in realtimeand is employed to direct illumination onto a tissue sample / specimen of thebreast region and to collect the photoacoustic signal therefrom. System 100 caninclude a photoacoustic probe 102 (also referred to as device 102, herein), pulsedlaser 104, oscilloscope 106, preamplifier 108 and computing device 110.

[0039] The photoacoustic probe 102 is shown in FIG. 1B can include an outercannula 112 (also referred to as cannula 112, herein), optical fiber cable 114, pulsedlaser 116, Lead-Zirconium-Titanium oxide (PZT) film detector 118 and quartzwindow 120 (also interchangeably referred to as quartz element 120, herein). Thephotoacoustic probe 102 can include the cannula 112 whose central cavity at thefront-sharp end is blocked by embedding quartz window 120. The quartz window120 is adapted to block any body fluid flowing into the cannula 112 throughcapillary action. The cannula 112 is blocked by the quartz window would house anoptical fiber cable 114 (also interchangeably referred to as hollow core fiber 114,herein) to carry the light for excitation. The PZT film detector 118 in contact withthe quartz window 120 at its dorsal end may serve as an acoustic sensor to detectphotoacoustic signals (also interchangeably referred to as a set of photoacousticsignals) induced in the tissue samples upon light excitation ex vivo / in vivo.

[0040] The photoacoustic spectra (also referred to as photoacoustic signals)from the samples were detected by the PZT film detector 118 and amplified by thepreamplifier 108. The time-domain, raw photoacoustic spectra were recorded in theoscilloscope 106 that is coupled to the computing device 110. Further, the dataanalysis on the acquired photoacoustic spectra can be performed on the computingdevice 110.

[0041] The device 102 can be fabricated with the outer cannula 112 for singleuse, disposable medical grade stainless steel cover with the embedded quartz 120at its tip, with the inner photoacoustic probe 102. Device 102 can be inserted intothe breast lump and can also be used on the surface of the residual cavity aftertumour excision to detect tumor at the margins.

[0042] In an exemplary embodiment, the excitation source 104, as presentedin the example, can be a pulsed laser, light-emitting diode (LED) or diode lasersource. The photoacoustic probe 102 can be targeted to the tissue specimen e.g.,tumour. The tissue specimen of the breast region can be illuminated from theLED / diode / laser source 104, and the corresponding photoacoustic spectra signalcan be collected through the same probe and recorded in the oscilloscope 106.

[0043] In an exemplary embodiment, the designed and developed probe 102can be standardized using ex vivo studies. To simulate the in vivo conditions, thephotoacoustic signals from the resected mastectomy specimen may be recordedusing ultrasound-guided probe insertion. The probe 102 may be targeted onto thetumor, the sample may be excited by 281 nm light from the LED / diode / laser source104, and the corresponding photoacoustic spectra signal can be collected throughthe same probe 102 and recorded in the oscilloscope 106.

[0044] The obtained photoacoustic signal can be pre-processed and analysedfor suitable features for differentiating the tissue types in the computing device 110.The computing device 110 can include a processor 202 operatively coupled to theoscilloscope 106. The raw, photoacoustic signals are detrended, baseline corrected,background subtracted, and normalized. A common region of interest (ROI) mayselect between 0.27 and 0.6 ms with a maximum variation for each photoacousticsignal.

[0045] The processor 202 shown in FIG. 2 is operatively coupled to a memory204, the memory 204 storing instructions executable by the processor 202 to receivethe set of photoacoustic signals from the oscilloscope 106. The processor 202 canpre-process and analyse the set of photoacoustic signals to extract a set of featuresfor differentiating the tissue types. The set of features pertaining to any or acombination of tissue structure and collagen patterns, where the tissue structure isfibroadenoma and phyllodes tissues. The pre-processed photoacoustic signals fromfibroadenoma and phyllodes tumor tissue samples can be subjected to a featureselection algorithm, 'minimal Redundancy Maximal Relevance' (mRMR) toeliminate unnecessary information and retain useful information. The mRMR usesthe mutual information-based empirical method to choose the features followed byfeature importance ranking value.

[0046] In the present disclosure, the top 20 features were optimal and can beconsidered as a feature matrix for the support vector machine learning algorithm.The processor 202 can classify the extracted set of features based on matching theextracted set of features with a reference set of features. The reference set offeatures can include stored information on tissue structure and collagen pattern ofthe tissue specimen. The classification pertaining to the set of tissue types of thebreast region, the set of tissue types pertaining to any or a combination of benign,pre-malignant and malignant state of the breast region, wherein based on acombination of classification of the extracted set of features, the processor isconfigured to determine a diagnosis for the breast region of the subject.

[0047] In an exemplary embodiment, the input feature matrix contained 20features belonging to each photoacoustic spectrum from fibroadenoma andphyllodes tumor. A multi-class Support Vector Machine learning model may betrained using 80% data and 20% for testing the trained model. In the presentdisclosure, SVM in radial basis function (RBF) kernel, to train and test the data byassessing the accuracy of the model for performance evaluation using the confusionmatrix.

[0048] The obtained photoacoustic spectra may be pre-processed andanalyzed for suitable features for differentiating the tissue types. A suitable machinelearning model can be developed to classify the spectral data based on the tissuetype. The spectra collected may be pre-processed and sent to a machine learningalgorithm for training and testing the model based on differential photoacousticspectra to classify tumor from normal.

[0049] The processor 202 is operatively coupled to the learning engine 210that is trained to classify the photoacoustic signal based on the tissue type toevaluate the benign, pre-malignant and malignant state of the tissue specimen. Thephotoacoustic signal collected can be pre-processed and sent to machine learningengine 210 to classify the photoacoustic signal based on tissue type as benign, premalignantand malignant within minutes.

[0050] Further, the machine learning engine 210 can be trained continuouslywith new cases and different tumour subtype models and updated periodically asversions to increase the specificity and sensitivity of the solution. A cloud-basedserver can be configured to update the repository periodically to maintain currentcases and data of the tissue specimen as versions to increase the specificity andsensitivity of the solution. The user can optionally update the learned model / versionupon paying royalty. The updating can be cloud-based, and the payment can beonline. The new user can avail of the latest, updated software version with thedevice.

[0051] FIG. 1B illustrates an exemplary photoacoustic probe design for thepurpose of breast lump diagnosis, in accordance with an embodiment of the presentdisclosure. The photoacoustic probe 102 has the cannula 112 to enclose the opticalfiber cable 114, wherein the optical fiber cable 114 adapted to direct illuminationfrom the excitation source 104 onto the breast region tissue specimen and collectthe set of photoacoustic signals therefrom. The cannula 112 used to insert the opticalfiber 114 into the tumour may be a single-use, disposable, medical-grade, stainlesssteel that is pre-sterilized and packed separately. The optical fiber 114 emergingfrom the excitation source 104 can be passed into the cannula 112 and can be reused.

[0052] The quartz element 120 embedded at the tip of the cannula 112restricts the flow of body fluid flowing through the cannula 112. The detector 118coupled to the quartz element 120 at the dorsal end detects the set of photoacousticsignals induced in the tissue specimen upon light excitation. The outer cannula 112is required to be disposable to fulfil the clinical requirements of ensuring sterilityand zero potential for transmitting disease from one patient to another. The innerphotoacoustic detector / probe 102 is reusable to reduce the cost of detection.

[0053] FIG. 1C illustrates intra-operative margin detection using thephotoacoustic probe, in accordance with an embodiment of the present disclosure.The intraoperative margin positivity detection after wide local excision duringbreast-conserving surgery, using in-house designed and developed photoacousticprobe 102 and its instrumentation followed by data analysis strategy for real-timediagnosis of the breast lump.

[0054] The present disclosure provides strong evidence for the applicabilityof photoacoustic spectroscopy and its capacity to efficiently discriminate minorchanges in protein expression levels in different tumors. The photoacoustic spectralsignatures generated by pulsed laser excitation at 281 nm from tissue fluorophorescan be used to predict in situ biochemical changes in different tissues. The presentdisclosure clearly demonstrates the potential of photoacoustic spectroscopy incancer detection / diagnosis. The photoacoustic spectroscopy utilizes non-ionizingradiations that may have strong translational and clinical significance.

[0055] The photoacoustic spectroscopy and machine learning detectmalignant residual tissues in the operated cavity and permit real-time confirmationof margin negative status in the residual cavity after Wide Local Excision (WLE).It may also permit the assessment of tumor on the surface / margins of the resectedWLE specimen.

[0056] Thus, the present invention overcomes the drawbacks, shortcomings,and limitations associated with existing solutions, and provides the device thatprovides a real-time tissue diagnosis and detection of margin positivity incomparison to obtaining a sample for frozen section evaluation or processing it byroutine histopathology, as is current practice. If margins are positive, it may requirere-excision, implying reoperation particularly if the facility of the frozen section isnot available.

[0057] FIG. 2 illustrates exemplary functional components 200 of theproposed system in accordance with an embodiment of the present disclosure.

[0058] In an aspect, the system 100 may comprise one or more processor(s)202. The one or more processor(s) 202 may be implemented as one or moremicroprocessors, microcomputers, microcontrollers, digital signal processors,central processing units, logic circuitries, and / or any devices that manipulate databased on operational instructions. Among other capabilities, the one or moreprocessor(s) 202 are configured to fetch and execute computer-readable instructionsstored in memory 204 of the system 100. The memory 204 may store one or morecomputer-readable instructions or routines, which may be fetched and executed tocreate or share the data units over a network service. The memory 204 maycomprise any non-transitory storage device including, for example, volatilememory such as RAM, or non-volatile memory such as EPROM, flash memory,and the like.

[0059] The system 100 may also comprise an interface(s) 206. Theinterface(s) 206 may comprise a variety of interfaces, for example, interfaces fordata input and output devices, referred to as I / O devices, storage devices, and thelike. The interface(s) 206 may facilitate communication of system 100. Theinterface(s) 206 may also provide a communication pathway for one or morecomponents of the system 100. Examples of such components include, but are notlimited to, processing engine(s) 208 and database 214.

[0060] The processing engine(s) 208 may be implemented as a combinationof hardware and programming (for example, programmable instructions) toimplement one or more functionalities of the processing engine(s) 208. In theexamples described herein, such combinations of hardware and programming maybe implemented in several different ways. For example, the programming for theprocessing engine(s) 208 may be processor-executable instructions stored on a nontransitorymachine-readable storage medium and the hardware for the processingengine(s) 208 may comprise a processing resource (for example, one or moreprocessors), to execute such instructions. In the present examples, the machinereadablestorage medium may store instructions that, when executed by theprocessing resource, implement the processing engine(s) 208. In such examples,system 100 may comprise the machine-readable storage medium storing theinstructions and the processing resource to execute the instructions, or the machinereadablestorage medium may be separate but accessible to device 100 and theprocessing resource. In other examples, the processing engine(s) 208 may beimplemented by electronic circuitry.

[0061] In an exemplary embodiment, the processing engine(s) 208 mayinclude learning engine 210, and other engine(s) 212. Database 214 may comprisedata that is either stored or generated as a result of functionalities implemented byany of the components of the processing engine(s) 208 or the system 100. Thelearning engine 210 is trained to evaluate the abnormal state of the tissue specimenin the breast region. The other engine(s) 212 can supplement the functionalities ofthe processing engine 208 or the system 100.EXPERIMENTAL RESULTS

[0062] In an experimental study, female patients above the age of 18 yearswith the diagnosis of a benign breast lump were included in the examination. Atotal of 30 samples were collected out of which 8 were phyllodes tumor and theremaining 22 were fibroadenoma. These 30 samples were subjected to aphotoacoustic spectroscopy-based study for further analysis.Sample collection:

[0063] The clearance from Institutional Ethics Committee was sought beforestarting the study. FNAC and core needle biopsy samples were collected from theex-vivo specimen. The samples were stored in Phosphate Buffer Solution (PBS)and appropriately labelled. The sample was frozen using liquid nitrogen and storedin a deep freezer at - 85 degrees in the Biophysics laboratory. The sample wasthawed to room temperature before the analysis.

[0064] Inclusion criteria- Female patients above 18 years who were admittedto the department of General Surgery with the diagnosis of a benign breast lumpwere included in the study.

[0065] Exclusion criteria- All the malignant tumor of the breast wereexcluded from the study.

[0066] The female patient with clinical features of breast lump may be seenby a consultant surgeon. The female patient may get admitted and surgicalintervention may be confirmed after which informed consent is obtained from thepatient by the Principal Investigator (PI) of the study. Only patients consenting tothe study were enrolled on the study. They were explained in the language theyunderstand.

[0067] After the specimen is removed by the consultant as a part of thestandard of care, FNAC and core needle samples were harvested without distortionof the specimen, thus making sure that future routine HPE analysis by thepathologist is not affected. The harvested samples were sent to the biophysicslaboratory for storage and later analysis was done using photoacousticspectrometry.

[0068] FIG. 3 illustrates an instrumentation setup used for obtainingphotoacoustic spectra from tissues at 281 nm pulsed laser excitations, in accordancewith an embodiment of the present disclosure.

[0069] In the photoacoustic spectral recording, all the tissue samples werethawed to room temperature. The photoacoustic signals were recorded from thetissue samples using the 281 nm excitation from Nd-YAG laser-pumped dye laserwith Rhodamine-6G dye. In the present disclosure ≈ 100 μJ energy / laser pulse isused, focusing the tissues under the study placed in the photoacoustic cell in contactmode with the detector. The instrumentation setup used to record the photoacousticsignals from tissues is shown in FIG. 3. The photoacoustic spectra from the sampleswere detected and amplified. The time-domain, raw photoacoustic spectra wererecorded in the oscilloscope. The photoacoustic signals recorded from each samplein different sites to obtain 5 spectra from each sample (Fibroadenoma - 5 spectra x25 tissues = 125 spectra; Phyllodes tumor = 5 spectra x 8 tissues = 40; Samplingfrequency - 2.6 MHz). Further, data analysis on the acquired photoacoustic spectracan be achieved using MATLAB R202a software.

[0070] The photoacoustic spectra acquired from the tissue samples such asfibroadenoma and phyllodes tumors in the region of 0.27 to 1.2 ms (at roomtemperature: 23 ± 2 °C) may be used for further data analysis.DATA ANALYSIS

[0071] FIG. 4A and FIG. 4B illustrate a graphical view representing the rawtime-domain photoacoustic spectra belonging to phyllodes and fibroadenoma, inaccordance with an embodiment of the present disclosure. The raw, time-domainphotoacoustic signals were processed for further analysis using MATLAB R2022asoftware and the typical photoacoustic signals of different tissue types in the region0-2 ms are shown in FIG. 4A and 4B respectively. The typical raw time-domainphotoacoustic spectra belonging to Phyllodes are shown in FIG. 4A andFibroadenoma are shown in FIG. 4B.PRE-PROCESSING

[0072] FIG. 5A and 5B illustrate a graphical view representing the typicalpre-processed photoacoustic spectra belonging to phyllodes and fibroadenoma, inaccordance with an embodiment of the present disclosure. The raw, photoacousticsignals were detrended, baseline corrected, background subtracted, and normalized.A common region of interest (ROI) was selected between 0.27 and 0.6 ms with amaximum variation for each photoacoustic signal FIG. 5A and 5B respectively. Thetypical pre-processed photoacoustic spectra belonging to Phyllodes are shown inFIG. 5A and Fibroadenoma are shown in FIG. 5B.

[0073] The differential protein expression in biological processes havereported the upregulation of certain proteins involved in amino acid metabolism,proteasome, fatty acid metabolism and glycolytic pathways. The differential proteinexpression in the tumor tissues would behave differently when excited with a pulsedlaser targeting the protein resulting in varied photoacoustic signatures compared tonormal tissues. Similarly, the varied protein expression in fibroadenoma andphyllodes tumor has shown differential photoacoustic signatures as shown in FIG.5A and 5B respectively.

[0074] The collagen plays a role in cancer fibrosis and is a prominentcomponent of the tumor microenvironment. Tumor cells can regulate collagenbiosynthesis influencing the tumor cell behavior. Microscopic changes in collagencomposition within tumor cells and matrix, play a role in the mutual feedback loopthat determines cancer prognosis, recurrence, and resistance.FEATURE SELECTION

[0075] FIG. 6 illustrates a graphical view representing the wavelet coefficientversus prediction rank values, in accordance with an embodiment of the presentdisclosure. FIG. 6 depicts typical photoacoustic spectra in the region of interest(ROI) from the fibroadenoma and phyllodes a time versus amplitude plot.

[0076] The pre-processed photoacoustic signals from fibroadenoma andphyllodes tumor tissue samples were subjected to the feature selection algorithm,mRMR to eliminate unnecessary information and retain useful information. ThemRMR uses the mutual information-based empirical method to choose the featuresfollowed by the feature importance ranking value shown in FIG. 6. In the presentdisclosure, the top 20 features were considered to be optimal and considered as afeature matrix for the support vector machine learning algorithm. The top 20features selected with their prediction rank values have been shown in table 1.Table 1: Top 20 features selected with their prediction rank value indescending order and their corresponding wavelet coefficient in the timedomain.CLASSIFICATION

[0077] FIG. 7 illustrates a schematic view of the confusion matrix showingthe performance of the support vector machine learning model, in accordance withan embodiment of the present disclosure. FIG. 7 depicts a confusion matrix showingthe performance of the support vector machine learning model for the classificationof fibroadenoma and phyllodes tumor with 96.9% accuracy.

[0078] The input feature matrix contained 20 features belonging to eachphotoacoustic spectrum from fibroadenoma and phyllodes tumor. A multi-classsupport vector machine learning model may be trained using 80% data and 20% fortesting the trained model. In the present disclosure, SVM is used in the radial basisfunction (RBF) kernel, to train and test the data by assessing the accuracy of themodel for performance evaluation using the confusion matrix shown in FIG. 7. Theconfusion matrix shows the performance of the support vector machine learningmodel for the classification of fibroadenoma and phyllodes tumor with 96.9%accuracy.

[0079] FIG. 8 illustrates a flow chart of a method for diagnosing breast lumpsin real-time using the photoacoustic probe, in accordance with an embodiment ofthe present disclosure.

[0080] Referring to FIG.8, method 800 includes at block 802 the processorcan receive the set of photoacoustic signals from the oscilloscope, the set ofphotoacoustic signals recorded in the oscilloscope detected by the detector. Thedetector 118 coupled to the quartz element at the dorsal end detects the set ofphotoacoustic signals induced in the tissue specimen upon light excitation. Thephotoacoustic probe has the cannula to enclose an optical fiber cable, where theoptical fiber cable is adapted to direct illumination from the excitation source ontothe breast region tissue specimen and collect the set of photoacoustic signalstherefrom. The quartz element embedded at the tip of the cannula restricts the flowof body fluid flowing through the cannula.

[0081] At block 804, the processor can pre-process and analyse the set ofphotoacoustic signals to extract the set of features. The set of features pertaining toany or a combination of fibroadenoma and phyllodes tissue. At block 806, theprocessor can classify the extracted set of features based on matching the extractedset of features with a reference set of features, the classification pertaining to a setof tissue types of the breast region, the set of tissue types pertaining to any or acombination of normal, benign, pre-malignant and malignant state of the breastregion. Based on a combination of classification of the extracted set of features, theprocessor is configured to determine a diagnosis for the breast region of the subject.

[0082] It will be apparent to those skilled in the art that the system 100 of thedisclosure may be provided using some or all of the mentioned features andcomponents without departing from the scope of the present disclosure. Whilevarious embodiments of the present disclosure have been illustrated and describedherein, it will be clear that the disclosure is not limited to these embodiments only.Numerous modifications, changes, variations, substitutions, and equivalents will beapparent to those skilled in the art, without departing from the spirit and scope ofthe disclosure, as described in the claims.ADVANTAGES OF THE PRESENT INVENTION

[0083] The present invention provides a device with a single-use, disposablemedical-grade needle for percutaneous insertion into a breast lump diagnosis anddetection of tumor positivity on the surface of the residual cavity after intraoperativetumor excision

[0084] The present invention provides a device that may provide a real-timetissue diagnosis and detection of margin positivity

[0085] The present invention provides a device with the integration ofmachine learning that avoids the essential requirement of a trained cytopathologistfor the evaluation of FNAC slides, enabling surgeons to diagnose breast pathologyindependently in clinics.

[0086] The present invention provides a device that updates machine-learningalgorithms upon the arrival of new breast tumour cases.

[0087] The present invention provides a device that provides cloud-basedtraining.

[0088] The present invention provides a device that avoids the requirementof external agents / chemicals for detection.

[0089] The present invention provides a device that increases the specificityand sensitivity of the solution.

[0090] The present invention provides a device that facilitates repeatassessments when clinically indicated, during follow up of patients.

[0091] The present invention provides a device that prevents unwantedsurgery for benign lumps alleviating the anxiety of the patients waiting for theFNAC results.

Claims

1. A system (100) for diagnosing tissue specimen of breast region of a subject, the system (100) comprising: a photoacoustic probe (102) having a cannula (112) to enclose an optical fiber cable (114), wherein the optical fiber cable (114) adapted to direct illumination from an excitation source (104) onto the tissue specimen of the breast region and to collect a set of photoacoustic signals therefrom; a quartz element (120) embedded at the tip of the cannula (112) restricts the flow of body fluid flowing through the cannula; a detector (118) coupled to the quartz element at dorsal end and detects the set of photoacoustic signals induced in the tissue specimen upon light excitation, the set of photoacoustic signals recorded in an oscilloscope (106); a processor (202) operatively coupled with a memory (204), the memory storing instructions executable by the processor to: receive the set of photoacoustic signals from the oscilloscope; preprocess and analyse the set of photoacoustic signals to extract a set of features, the set of features pertaining to any or a combination of fibroadenoma and phyllodes tissue; and classify the extracted set of features based on matching the extracted set of features with a reference set of features, the classification pertaining to a set of tissue types of the breast region, the set of tissue types pertaining to any or a combination of normal, benign, pre-malignant and malignant state of the breast region, wherein based on a combination of classification of the extracted set of features, the processor is configured to determine a diagnosis for the breast region of the subject.

2. The system as claimed in claim 1, wherein the optical fiber cable (114) emerging from the excitation source (104) is passed into the cannula (112) to direct illumination into tissue specimen of the breast region.

3. The system as claimed in claim 1, wherein the set of photoacoustic signals collected through the photoacoustic probe (102) and recorded in the oscilloscope (106), the oscilloscope (106) coupled to the excitation source.

4. The system as claimed in claim 1, wherein the set of photoacoustic signals from fibroadenoma and phyllodes tissues are subjected to a feature selection algorithm to eliminate unnecessary information and retain useful information.

5. The system as claimed in claim 1, wherein the feature selection algorithm uses a mutual information-based empirical approach to select the features followed by feature importance ranking value.

6. The system as claimed in claim 1, wherein the processor (202) is operatively coupled to a learning engine (210), the learning engine (210) adapted to receive the pre-processed set of features and trained to evaluate the set of tissue types as normal, benign, pre-malignant and malignant state of the breast region.

7. The system as claimed in claim 1, wherein the photoacoustic probe (102) detects malignant residual tissues in the operated cavity and permits realtime confirmation of margin negative status in the residual cavity after wide local excision (WLE), the photoacoustic probe permits the assessment of tumor on the margins of the resected WLE specimen.

8. The system as claimed in claim 1, wherein the processor (202) is operatively coupled to a cloud-based server that is configured to update the database periodically to maintain current cases and data of the tissue specimen.

9. A photoacoustic probe (102) for diagnosing tissue specimen of breast region of a subject, the photoacoustic probe comprising: a cannula (112) to enclose an optical fiber cable (114), wherein the optical fiber cable (114) adapted to direct illumination from an excitation source (104) onto the tissue specimen of the breast region and to collect a set of photoacoustic signals therefrom; a quartz element (120) embedded at the tip of the cannula (112) restricts the flow of body fluid flowing through the cannula; a detector (118) coupled to the quartz element at dorsal end detects the set of photoacoustic signals induced in the tissue specimen upon light excitation, the set of photoacoustic signals recorded in an oscilloscope (106); a processor (202) operatively coupled with a memory (204), the memory storing instructions executable by the processor to: receive the set of photoacoustic signals from the oscilloscope; preprocess and analyse the set of photoacoustic signals to extract a set of features, the set of features pertaining to any or a combination of fibroadenoma and phyllodes tissue; and classify the extracted set of features based on matching the extracted set of features with a reference set of features, the classification pertaining to a set of tissue types of the breast region, the set of tissue types pertaining to any or a combination of normal, benign, pre-malignant and malignant state of the breast region, wherein based on a combination of classification of the extracted set of features, the processor is configured to determine a diagnosis for the breast region of the subject.

10. A method (800) for diagnosing tissue specimen of breast region of a subject, the method comprising: receiving (802), at a processor, a set of photoacoustic signals from a oscilloscope, the set of photoacoustic signals recorded in the oscilloscope detected by a detector that is operatively coupled to a quartz element at dorsal end to detect the set of photoacoustic signals induced in the tissue specimen upon light excitation, wherein a photoacoustic probe (102) having a cannula (112) to enclose an optical fiber cable (114), wherein the optical fiber cable (114) adapted to direct illumination from an excitation source (104) onto the tissue specimen of the breast region and to collect the set of photoacoustic signals therefrom; pre-processing and analysing (804), at the processor, the set of photoacoustic signals to extract a set of features from the set of photoacoustic signals, the set of features pertaining to any or a combination of fibroadenoma and phyllodes tissue; and classifying (806), at the processor, the extracted set of features based on matching of the extracted set of features with a reference set of features, the classification pertaining to a set of tissue types of the breast region, the set of tissue types pertaining to any or a combination of normal, benign, premalignant and malignant state of the breast region, wherein based on a combination of classification of the extracted set of features, the processor is configured to determine a diagnosis for the breast region of the subject.