Multi-device combined use-based identification method and system for raw materials and excipients in formulation

Abstract

A multi-device combined use-based identification method and system for raw materials and excipients in a formulation. The method comprises: performing cryogenic polishing on a cross-section of a test sample under analysis to obtain a cross-section under test; using an electron microscope to observe the cross-section under test to obtain an electron microscopy image of the cross-section under test, and determining a region under test on the cross-section under test; using an energy dispersive spectrometer to scan the region under test to obtain an energy spectrum image; on the basis of the electron microscopy image and the energy spectrum image, determining whether there are characteristic elements in the test sample, and if there are characteristic elements, identifying the characteristic elements on the basis of the energy spectrum image; and on the basis of the electron microscopy image and the energy spectrum image, determining whether there are components matching preset morphological characteristics in the test sample, and if there are components matching the preset morphological characteristics, using a pre-constructed convolutional neural network model to perform analysis processing on the electron microscopy image to identify the components matching the preset morphological characteristics. The identification method can more accurately and quickly identify components in a sample under analysis.

Description

A method and system for identifying raw materials and excipients in pharmaceutical preparations based on multi-device integration

[0001] This disclosure claims priority to Chinese Patent Application No. CN202411913446.3, filed on December 24, 2024, entitled "A Method and System for Identifying Raw Materials and Excipients in a Formulation Based on Multi-device Co-operation", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This invention relates to the field of formulation analysis technology, and in particular to a method and system for identifying raw materials and excipients in formulations based on the use of multiple devices. Background Technology

[0003] In the field of pharmaceutical analysis, traditional chromatographic and mass spectrometric methods hold a crucial position. High-performance liquid chromatography (HPLC) and gas chromatography (GC) utilize the interaction between the sample and the stationary phase to separate components, effectively handling complex mixtures and helping researchers quantitatively analyze the purity and concentration of drug components. Meanwhile, mass spectrometry provides molecular weight and structural information by ionizing the sample and measuring the mass-to-charge ratio of ions. Typically, these two techniques are used in combination (such as LC-MS or GC-MS) to accurately identify and quantify low-concentration components in complex matrices. However, chromatographic and mass spectrometric analyses usually require sample extraction or dissolution, which may damage some sensitive samples, and the analysis time is relatively long, thus limiting their suitability for rapid and non-destructive analysis.

[0004] The above background information is provided only to assist in understanding the inventive concept and technical solution of this invention. It does not necessarily belong to the prior art of this application, nor does it necessarily provide technical teaching. In the absence of clear evidence that the above information was disclosed before the filing date of this application, the above background information should not be used to evaluate the novelty and inventiveness of this application. Summary of the Invention

[0005] The purpose of this invention is to provide a method and monitoring system for identifying raw materials and excipients in pharmaceutical preparations based on the use of multiple devices, which can achieve more accurate and faster identification of components in the sample to be analyzed.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for identifying raw materials and excipients in a pharmaceutical preparation based on the use of multiple devices includes the following steps:

[0008] A cross section of the test sample to be analyzed is subjected to cryo-polishing to obtain the test section;

[0009] The cross section to be tested is observed using an electron microscope to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested;

[0010] The test area was scanned using an energy dispersive spectrometer to obtain an energy spectrum image;

[0011] The electron microscope image and the energy dispersive spectroscopy image are used to determine whether the test sample contains characteristic elements. If characteristic elements are present, the characteristic elements are identified based on the energy dispersive spectroscopy image.

[0012] Based on the electron microscope image and the energy dispersive spectroscopy image, it is determined whether the test sample contains components that match the preset morphological features. If there are components that match the preset morphological features, the electron microscope image is analyzed and processed using a pre-built convolutional neural network model to identify the components that match the preset morphological features.

[0013] Furthermore, following any one or a combination of the aforementioned technical solutions, the method further includes the following steps:

[0014] If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area.

[0015] The composition of the test sample was obtained based on the Raman spectral analysis.

[0016] Furthermore, following any one or a combination of the aforementioned technical solutions, in the process of analyzing the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range.

[0017] The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

[0018] Furthermore, following any one or a combination of the aforementioned technical solutions, the method further includes using a pre-built deep learning model to analyze and process the energy spectrum image to identify the feature elements. The deep learning model is configured to perform feature extraction, noise reduction, and image enhancement processing on the energy spectrum image to identify and display the feature elements.

[0019] Furthermore, following any one or a combination of the aforementioned technical solutions, the electron microscope image is analyzed and processed using a pre-built convolutional neural network model to identify components that match preset morphological features, including:

[0020] The electron microscope image is preprocessed to obtain a processed electron microscope image. The preprocessing includes one or more of denoising, normalization, and size adjustment.

[0021] The processed electron microscope image is input into the convolutional neural network model. The convolutional neural network model predicts and segments the components of the processed electron microscope image that match the preset morphological features based on the morphological features of various polymer compounds that have been pre-annotated and learned.

[0022] Furthermore, following any one or a combination of the aforementioned technical solutions, the convolutional neural network model is pre-constructed in the following manner:

[0023] A learning sample set is obtained, which includes multiple learning samples. Each learning sample includes one or more electron microscope image samples corresponding to a known component. The known component has regular morphological features, and each electron microscope image sample is labeled with its corresponding known component.

[0024] A basic neural network model is determined, the learning sample set is input into the basic neural network model, and the basic neural network model is trained using a preset loss function to obtain the convolutional neural network model.

[0025] Furthermore, following any one or a combination of the aforementioned technical solutions, the components of the pre-defined morphological features in the processed electron microscope image are predicted and segmented to determine the components of each pre-defined morphological feature, including:

[0026] Determine the substances corresponding to the components of each pre-defined morphological feature;

[0027] Obtain the segmented images corresponding to the components of each preset morphological feature. Each segmented image is configured to display a component of a preset morphological feature.

[0028] Furthermore, following any one or a combination of the aforementioned technical solutions, it also includes:

[0029] The electron microscope image of the test section obtained by observing the test section with an electron microscope is configured as the first electron microscope image;

[0030] Based on the electron microscope image and the energy dispersive spectroscopy image, determine whether there are morphological features in the test sample. If morphological features are present, analyze and process the electron microscope image using a pre-built convolutional neural network model to identify components that match preset morphological features, including the following steps:

[0031] The test area is observed using an electron microscope to obtain a second electron microscope image of the test area, wherein the resolution of the second electron microscope image is greater than that of the first electron microscope image;

[0032] The second electron microscope image is then analyzed and processed using a pre-built convolutional neural network model to identify components that match preset morphological features.

[0033] Furthermore, following any one or a combination of the aforementioned technical solutions, for different components in the test sample that match morphological features, after analyzing and processing the electron microscope image using a pre-built convolutional neural network model to identify components matching preset morphological features, the method further includes:

[0034] The energy spectrum image is further analyzed and processed using a pre-built deep learning model to identify components with similar morphological features.

[0035] Furthermore, based on any one or a combination of the aforementioned technical solutions, the presence of a component matching a preset morphological characteristic in the test sample is determined using the following method:

[0036] A morphological feature component library is constructed by pre-determining multiple components with regular morphological features. The morphological feature component library includes multiple components and reference morphological features corresponding to each component. The reference morphological features can be one or more.

[0037] The electron microscope image and the energy dispersive spectroscopy image are analyzed and compared with the morphological feature component library. If the morphological features of the component to be analyzed in the electron microscope image and / or the energy dispersive spectroscopy image match the reference morphological features, then it is determined that the test sample has a component with a matching preset morphological features.

[0038] Furthermore, based on any one or a combination of the aforementioned technical solutions, the analytical objects applicable to the multi-device-based formulation raw material and excipient identification method include at least: formulations, polymeric materials, and semiconductor materials; and / or,

[0039] It also includes the following steps:

[0040] If the test sample contains neither characteristic elements nor components matching the preset morphological characteristics, then X-ray diffraction analysis is used to analyze the composition of the test sample.

[0041] According to another aspect of the present invention, the present invention provides a formulation raw material and excipient identification system based on multi-device combination, including an electron microscope, an energy dispersive spectroscopy instrument and a processor;

[0042] The formulation raw material and excipient identification system is configured to identify raw materials and excipients in the formulation in the following ways:

[0043] A cross section of the test sample to be analyzed is cryopolished to obtain the test section;

[0044] The electron microscope is used to observe the cross section to be tested to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested;

[0045] The energy dispersive spectrometer is used to scan the area to be tested to obtain an energy spectrum image;

[0046] The processor determines whether the test sample contains characteristic elements based on the electron microscope image and the energy dispersive spectroscopy image. If characteristic elements are present, the processor identifies the characteristic elements based on the energy dispersive spectroscopy image.

[0047] The processor determines whether the test sample contains components that match preset morphological features based on the electron microscope image and the energy dispersive spectroscopy image. If there are components that match preset morphological features, the processor uses a pre-built convolutional neural network model to analyze and process the electron microscope image to identify the components that match preset morphological features.

[0048] Furthermore, any one or a combination of the aforementioned technical solutions also includes a Raman spectrometer;

[0049] The formulation raw material and excipient identification system is also configured to identify raw materials and excipients in the formulation in the following ways:

[0050] If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area.

[0051] The processor obtains the composition of the test sample based on the Raman spectrum analysis.

[0052] Furthermore, following any one or a combination of the aforementioned technical solutions, it also includes an atomic force microscope;

[0053] The formulation raw material and excipient identification system is also configured to identify raw materials and excipients in the formulation in the following ways:

[0054] In the process of analyzing the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range.

[0055] The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

[0056] The beneficial effects of the technical solution provided by this invention are as follows:

[0057] a. This invention can greatly shorten the analysis time of each finished product of the sample by selecting electron microscopy, energy dispersive spectroscopy or convolutional neural network model to analyze components with characteristic morphology for different components in the sample to be tested, and obtain more accurate, larger range and higher resolution analysis results in a shorter time.

[0058] b. This invention targets different components in a sample that have matching / similar / close morphological features, and further combines energy dispersive spectroscopy (EDS) to effectively distinguish different components with similar morphological features, thus avoiding errors or omissions in the analysis of components with similar morphological features;

[0059] c. The formulation raw material and excipient identification method based on multi-device combination proposed in this invention is not only applicable to the identification of formulation raw materials and excipients, but also to the identification of polymer materials and semiconductor materials. It can be applied to a variety of industries such as polymer materials and semiconductors, and has a wide range of applications. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 is a general flowchart of a method for identifying raw and excipient materials in a formulation provided by an exemplary embodiment of the present invention;

[0062] Figure 2 is a partial flowchart of a method for identifying raw and excipient materials in a formulation provided by an exemplary embodiment of the present invention.

[0063] Figure 3 is a first electron micrograph of a sample to be tested provided in an exemplary embodiment of the present invention;

[0064] Figure 4 is a second electron microscope image of the sample to be tested provided in an exemplary embodiment of the present invention;

[0065] Figure 5 is an energy spectrum image of the S element in the sample to be tested provided in an exemplary embodiment of the present invention;

[0066] Figure 6 is an energy spectrum image of Mg in a sample to be tested provided in an exemplary embodiment of the present invention;

[0067] Figure 7 is an image showing the distribution of S element in a test sample provided by an exemplary embodiment of the present invention;

[0068] Figure 8 is an image showing the distribution of Mg elements in a test sample provided by an exemplary embodiment of the present invention;

[0069] Figure 9 is a reference image showing the morphological characteristics of corn starch in a formulation according to an exemplary embodiment of the present invention;

[0070] Figure 10 is a reference image showing the morphological characteristics of low-substituted hydroxypropyl cellulose in a formulation according to an exemplary embodiment of the present invention;

[0071] Figure 11 is a reference image showing the morphological characteristics of lactose in a formulation according to an exemplary embodiment of the present invention;

[0072] Figure 12 is a segmentation image of corn starch in a birepiperazole sample provided in an exemplary embodiment of the present invention;

[0073] Figure 13 is a segmentation image of low-substituted hydroxypropyl cellulose in a birepiperazole sample provided in an exemplary embodiment of the present invention;

[0074] Figure 14 is a segmentation image of lactose in a birepiperazole sample provided in an exemplary embodiment of the present invention;

[0075] Figure 15 is a schematic diagram of a microcrystalline candidate region to be Raman-confirmed according to an exemplary embodiment of the present invention;

[0076] Figure 16 is a Raman spectrum of a substance at a C element collection point provided in an exemplary embodiment of the present invention;

[0077] Figure 17 is a graph showing the analytical results of a birepiperazole sample provided in an exemplary embodiment of the present invention;

[0078] Figure 18 is a comparison of the analytical results of the present invention and the Raman method on the same sample provided by an exemplary embodiment of the present invention. Detailed Implementation

[0079] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0080] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0081] In one embodiment of the present invention, a method for identifying raw materials and excipients in a formulation based on the use of multiple devices is provided. Referring to Figure 1, the method includes the following steps:

[0082] A cross section of the test sample to be analyzed is subjected to cryo-polishing to obtain the test section;

[0083] The cross section to be tested is observed using an electron microscope to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested;

[0084] The test area was scanned using an energy dispersive spectrometer to obtain an energy spectrum image;

[0085] The electron microscope image and the energy dispersive spectroscopy image are used to determine whether the test sample contains characteristic elements. If characteristic elements are present, the characteristic elements are identified based on the energy dispersive spectroscopy image.

[0086] Based on the electron microscope image and the energy dispersive spectroscopy image, it is determined whether the test sample contains components that match the preset morphological features. If there are components that match the preset morphological features, the electron microscope image is analyzed and processed using a pre-built convolutional neural network model to identify the components that match the preset morphological features.

[0087] There are several ways to determine whether the test sample contains components that match the preset morphological characteristics. One way is to use manual judgment based on experience; another way is to use automatic judgment through the following methods:

[0088] A morphological feature component library is constructed by pre-determining multiple components with regular morphological features. The morphological feature component library includes multiple components and reference morphological features corresponding to each component. The reference morphological features can be one or more.

[0089] The electron microscope image and the energy dispersive spectroscopy image are analyzed and compared with the morphological feature component library. If the morphological features of the component to be analyzed in the electron microscope image and / or the energy dispersive spectroscopy image match the reference morphological features, then it is determined that the test sample has a component with a matching preset morphological features.

[0090] Preferably, the analysis method for components with pre-defined matching morphological features (also known as components with morphological features) includes the following steps:

[0091] The electron microscope image is preprocessed to obtain a processed electron microscope image. The preprocessing includes one or more of denoising, normalization, and size adjustment.

[0092] The processed electron microscope image is input into the convolutional neural network model. Based on the morphological features of various polymer compounds learned through pre-annotation, the convolutional neural network model predicts and segments the components in the processed electron microscope image that match preset morphological features to determine the components of each preset morphological feature. This includes: determining the substances corresponding to the components of each preset morphological feature; and acquiring segmented images corresponding to the components of each preset morphological feature, wherein each segmented image is configured to display a component that matches a preset morphological feature.

[0093] In one embodiment of the present invention, the convolutional neural network model is pre-constructed in the following manner:

[0094] A learning sample set is obtained, which includes multiple learning samples. Each learning sample includes one or more electron microscope image samples corresponding to a known component. The known component has regular morphological features, and each electron microscope image sample is labeled with its corresponding known component.

[0095] A basic neural network model is determined, the learning sample set is input into the basic neural network model, and the basic neural network model is trained using a preset loss function to obtain the convolutional neural network model.

[0096] Regarding the test sample containing different components with matching / similar / close morphological features, it should be noted that, in this application, different components with matching / similar / close morphological features refer to two different components that have similar but not completely identical morphological features. Specifically, the two components have the same shape of their morphological contours but differ in size, or the two components have similar sizes (size difference within a certain range), and the morphological contours of the two components are not significantly different. For example, if the degree of matching of the morphological contours of the two components is greater than a certain first preset value (e.g., 80% or 90%), then the two components are judged to have matching / similar / close morphological features. In this case, if only a pre-constructed convolutional neural network model is used to analyze and process the electron microscope image to identify the components that match the preset morphological features, errors or omissions may occur.

[0097] Therefore, in one embodiment of the present invention, for different components in the test sample that have matching / similar / close morphological features, after analyzing and processing the electron microscopy image using a pre-built convolutional neural network model to identify components that match preset morphological features, the method further includes:

[0098] The energy dispersive spectroscopy (EDS) images are further analyzed using a pre-built deep learning model to identify components with similar morphological features. Thus, by observing the elemental composition of different regions through EDS, components with similar morphological features in a sample can be efficiently and accurately distinguished. For example, some particles may be very similar in morphology, but their constituent elements may differ, or the distribution of elements may vary in different regions. By comparing these elemental distributions, subtle differences between substances can be discovered, helping to determine whether other unique elements are present, or whether different elemental distribution patterns exist, thereby further confirming the type and structure of the compound.

[0099] In the combined electron microscopy (EMS) and energy dispersive spectroscopy (EDS) analysis, high-resolution images of the sample are acquired using an EMS to observe its microstructure. During this process, sample preparation and imaging conditions are optimized to obtain clear and detailed images. Next, the acquired EMS images undergo denoising and enhancement processing to improve overall image quality. This step includes techniques such as filtering and histogram equalization to enhance the visual features of the sample, making the analysis more accurate. Subsequently, the sample is analyzed using an EDS to obtain elemental distribution information. Through analysis, the relative abundance of each element can be obtained, and corresponding elemental mapping maps are generated to better understand the sample composition. At this stage, the EMS images and EDS data are registered to analyze the distribution of elements in the microstructure. Ensuring accurate registration of the two data sets in the same coordinate system is crucial for effective joint analysis. Finally, the EMS images and EDS data are combined to generate a detailed analysis report summarizing the sample's microstructure and elemental composition. This report provides important evidence for subsequent research, helping to further explore the sample's characteristics and applications.

[0100] In this embodiment, the method further includes the following steps:

[0101] If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area.

[0102] The composition of the test sample was obtained based on the Raman spectral analysis.

[0103] In the absence of obvious morphological features and characteristic elements, Raman spectroscopy is an effective non-destructive testing technique. By acquiring Raman spectra of a target region, a Raman spectrum of that region is obtained. Each compound possesses unique molecular vibrational characteristics, manifested as specific Raman scattering peaks, which can be used to distinguish different compounds. By comparing the obtained spectra with an established Raman database, the compound type of the sample can be accurately determined. This method is particularly suitable for analyzing complex samples containing multiple substances and can provide high-resolution molecular information. However, Raman spectroscopy analysis of sample composition is time-consuming; if each component of the test sample is analyzed using Raman spectroscopy, the time required will be considerable. Referring to Figure 18, in a specific embodiment of the present invention, the method provided by the present invention, based on a multi-device combined approach for identifying raw materials and excipients in a formulation, as shown on the left side of Figure 18, utilizes a combined analysis method of electron microscopy + energy dispersive spectroscopy + Raman spectroscopy to determine the components of the sample to be tested in just 2 hours. Compared to the existing method of using Raman spectroscopy to analyze the components of the sample to be tested, which requires 13 hours (as shown on the right side of Figure 18), this method significantly shortens the analysis time. Furthermore, this method, through 2 hours of testing and analysis, obtains a 10mm... 2 The analysis results, obtained with a test range and 300nm resolution, are comparable to those obtained using the existing single Raman method at 5mm. 2 The analysis results with a test range and 11μm resolution show that this method not only greatly shortens the analysis time, but also obtains analysis results with a larger test range and higher resolution.

[0104] Electron microscopy provides nanometer-level spatial resolution, enabling researchers to observe the microstructure and particle morphology of samples in depth, thereby effectively understanding the particle distribution and interactions in pharmaceutical formulations. Simultaneously, energy dispersive spectroscopy (EDS) can acquire elemental composition information of samples in real time, providing a comprehensive understanding of the chemical composition and distribution of materials. The proposed multi-device method for identifying raw materials and excipients within pharmaceutical formulations, combining electron microscopy, EDS, and Raman spectroscopy, demonstrates significant advantages in formulation research. Compared to traditional chromatography and mass spectrometry methods, this combined approach not only reduces the complexity of sample processing but also allows for non-destructive analysis. In particular, the application of Raman spectroscopy enables researchers to obtain results rapidly without damaging the sample. This speed is especially suitable for online monitoring and quality control. The comprehensive use of these technologies can reveal the characteristics of formulations more comprehensively, providing a richer perspective for in-depth analysis.

[0105] The shortcomings of existing methods are also reflected in their technical limitations and reliance on a single technique. Traditional analytical methods typically rely on a single technique, such as chemical analysis, spectroscopic analysis, or morphological analysis, but these methods often fail to comprehensively and accurately capture all the information of complex samples. For example, some substances may lack both significant elemental characteristics and obvious morphological features, and existing methods are often insufficient to extract valuable data from such substances. Existing methods typically select analytical means based on known material properties, lacking flexibility and failing to automatically adjust analytical strategies according to the actual characteristics of the sample. This fixed approach often fails to provide the most suitable analytical path when dealing with different types of substances. Especially for substances lacking both characteristic elements and morphological features, existing technologies may not provide effective analytical means, leading to inaccurate or incomplete analytical results. Existing methods are inefficient and often unsatisfactory in processing complex samples. Traditional image processing or simple chemical analysis techniques, especially when dealing with complex substances such as high-molecular-weight hydrocarbons and microcrystalline cellulose, are prone to misjudgments or excessively long analysis times, affecting data reliability and analytical efficiency. In the analysis of multi-component samples, existing methods struggle to accurately distinguish and quantify each component, especially when there is overlap between substances, making information extraction and identification quite difficult.

[0106] The formulation raw material and excipient identification method proposed in this invention based on multi-device combination can effectively solve the problems of existing methods relying on single analysis technology, lack of flexibility and accuracy. When processing complex or multi-component samples, it can extract key component information of the sample to be analyzed more efficiently and accurately.

[0107] In one embodiment of the present invention, referring to Figure 2, during the analysis of the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range.

[0108] The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

[0109] Atomic force microscopy (AFM) scans the surface morphology of samples to obtain their microstructure information. While primarily used for surface morphology analysis, in some cases, combining it with analysis of mechanical properties (such as nanomechanics and friction) can help distinguish different compounds. By comparing the mechanical responses of different substances, the composition of the sample can be inferred. Although this method cannot directly provide all chemical composition information, it can, in certain situations, help further screen compounds and provide clues for subsequent analysis.

[0110] It should be noted that if the test sample contains neither characteristic elements nor components matching the preset morphological characteristics, in addition to using X-ray diffraction (XRD) analysis to determine the composition of the test sample, XRD analysis can also be used to determine the composition of the test sample. X-ray diffraction (XRD) is a powerful materials analysis technique, mainly used to analyze the crystal structure of substances. Each compound or crystal has a unique fingerprint-like characteristic in its diffraction pattern. By measuring the diffraction intensity of the sample at different angles, the crystal structure information of the compound can be obtained. Although XRD is more suitable for crystalline materials, it can provide information such as the lattice constant and interplanar spacing of a substance, helping to identify the type of substance. By comparing known diffraction patterns in a database, the chemical composition of the sample can be inferred, thereby achieving compound identification.

[0111] In a specific embodiment of the present invention, a method for rapidly analyzing the composition and distribution of irinotecan is provided, specifically including the following steps:

[0112] (1) Take a birepiperazole test sample and perform cryopolishing treatment on the test sample along the maximum cross section using a soft ion beam cryopolisher. Then, observe the cross section of the test sample after polishing using an electron microscope to obtain a complete first electron microscope image of the cross section as shown in Figure 3, and determine a test area in the cross section as the rectangular box in Figure 3.

[0113] (2) The formulation of the buripiperazole sample consists of buripiperazole, magnesium stearate, lactose, low-substituted hydroxypropyl cellulose, corn starch, and microcrystalline cellulose.

[0114] In formulating subsequent processing methods, this application will conduct the following analysis based on the above prescription:

[0115] For birepiperazole and magnesium stearate, which have characteristic elements, characteristic element analysis is used to extract effective information.

[0116] For high-molecular-weight hydrocarbons such as lactose, low-substituted hydroxypropyl cellulose, and corn starch that have morphological characteristics but lack characteristic elements, deep learning networks will be used for processing.

[0117] For microcrystalline cellulose that lacks both morphological and characteristic elemental characteristics, Raman spectroscopy is employed for analysis to obtain more comprehensive material information. Existing methods may not select the most suitable analytical technique based on different material properties (such as the presence or absence of characteristic elements or morphological features). Our method, however, through precise classification, selects appropriate analytical techniques (elemental analysis, deep learning, Raman spectroscopy) for different properties, making it more flexible and efficient. For substances lacking characteristic elements but possessing morphological characteristics, existing methods may rely on traditional image processing or chemical analysis techniques, while our method utilizes deep learning networks to learn morphological features from data, resulting in greater intelligence and accuracy. For substances lacking both morphological and characteristic elemental characteristics (such as microcrystalline cellulose), existing methods may struggle to analyze them effectively; our innovative method employs Raman spectroscopy to supplement the analysis, enabling the acquisition of more comprehensive information.

[0118] (3) Among them, buriperazole and magnesium stearate have characteristic elements S and Mg.

[0119] The characteristic elements typically refer to chemical elements unique to certain substances or components (substances not found in other components), which serve as markers for identification and analysis. In pharmaceutical or material analysis, characteristic elements are crucial for distinguishing or identifying active ingredients (such as the main components of a drug) or excipients (such as additives or fillers). For example, if a certain active ingredient or excipient in a formulation contains iron (Fe), while other components do not, then iron can be considered a characteristic element of that active ingredient. By detecting these characteristic elements, we can accurately locate or identify different components in a drug.

[0120] Under an electron microscope, the region to be tested was scanned using an energy dispersive spectroscopy (EDS) instrument used in conjunction with the electron microscope to obtain a second electron microscope image of the region, as shown in Figure 4. The resolution of the second electron microscope image is greater than that of the first electron microscope image, and the distribution of all characteristic elements in the region is obtained, including the S element in the selected region shown in Figure 5 and the Mg element in the selected region shown in Figure 6.

[0121] The energy spectrum image is analyzed and processed using a pre-built deep learning model to identify the feature elements. Specifically, features are extracted from its neighborhood (e.g., pixels within a local window) through a multi-layer convolutional neural network. The energy spectrum image is then subjected to feature extraction, and noise is effectively removed by utilizing the structural information of the image itself. Finally, image enhancement techniques are used to generate the S element distribution map shown in Figure 7 and the Mg element distribution map shown in Figure 8.

[0122] (4) Among them, lactose, low-substituted hydroxypropyl cellulose, and corn starch are high-molecular-weight compounds with morphological characteristics but lacking characteristic elements. The samples were imaged using a high-resolution electron microscope to obtain a high-precision image of the sample surface morphology, namely the second electron microscope image, as shown in Figure 2.

[0123] The acquired second electron microscopy images were preprocessed, including denoising, normalization, and resizing. A pre-trained model was loaded, which had learned the morphological features of carbon, hydrogen, and oxygen polymers based on labeled data. Figures 9 to 11 are reference images of the morphological features of corn starch, low-substituted hydroxypropyl cellulose, and lactose in the formulation, respectively.

[0124] The preprocessed second electron microscope image is input into the trained model for prediction and segmentation. The prediction and segmentation results are output in the form of data and images, as shown in Figures 12 to 14. Figure 12 is the segmentation image of corn starch, Figure 13 is the segmentation image of low-substituted hydroxypropyl cellulose, and Figure 14 is the segmentation image of lactose.

[0125] (5) For microcrystalline cellulose that has neither morphological characteristics nor characteristic elements, Raman spectroscopy is used for analysis. Within the region selected in step (3), the Raman spectrum of the substance at the collection point is detected using a confocal micro-Raman imaging system at the same coordinates to obtain the Raman spectrum of that region. Each compound has unique molecular vibrational characteristics, manifested as specific Raman scattering peaks, which can be used to distinguish different compounds. By comparing the obtained spectrum with the established Raman database, the type of compound the sample belongs to can be accurately determined. Figure 16 shows the Raman spectrum of the substance at the C element collection point (Figure 15), corresponding to microcrystalline cellulose.

[0126] (6) Cases of identical characteristic elements: If different compounds or components contain the same characteristic element, the concentration and distribution of that element can be assessed in detail using energy dispersive X-ray spectroscopy (EDS). EDS provides information on the local distribution and concentration of elements. By observing the distribution of elements in different regions, the role of that element in the sample can be inferred. For example, active ingredients typically have higher concentrations in pharmaceutical preparations and may be concentrated in specific areas, while excipients may exhibit a more uniform distribution. By analyzing the differences in element concentration and distribution patterns, different components can be effectively distinguished, and the chemical composition of the sample can be further confirmed.

[0127] (7) Figure 17 shows the final analysis results. The colors of different regions represent different active ingredients. This figure can be used to determine the content, particle size and other data of different active ingredients.

[0128] The formulation raw material and excipient identification method based on multi-device combination provided by the present invention is not only applicable to the identification of formulation raw materials and excipients, but also to the identification of polymer materials and semiconductor materials. It can be applied to a variety of industries such as polymer materials and semiconductors, and has a wide range of applications.

[0129] In one embodiment of the present invention, a formulation raw material and excipient identification system based on multi-device combination is provided, including an electron microscope, an energy dispersive spectroscopy (EDS) instrument, a Raman spectrometer, an atomic force microscope (AFM) instrument, and a processor. It can correlate electron microscopy, EDS, Raman, and AFM data to analyze the sample to be analyzed, and identify the components of the sample to be analyzed more accurately and quickly.

[0130] The formulation raw material and excipient identification system is configured to identify raw materials and excipients in the formulation in the following ways:

[0131] A cross section of the test sample to be analyzed is cryopolished to obtain the test section;

[0132] The electron microscope is used to observe the cross section to be tested to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested;

[0133] The energy dispersive spectrometer is used to scan the area to be tested to obtain an energy spectrum image;

[0134] The processor determines whether the test sample contains characteristic elements based on the electron microscope image and the energy dispersive spectroscopy image. If characteristic elements are present, the processor identifies the characteristic elements based on the energy dispersive spectroscopy image.

[0135] The processor determines whether the test sample contains components that match preset morphological features based on the electron microscope image and the energy dispersive spectroscopy image. If there are components that match preset morphological features, the processor uses a pre-built convolutional neural network model to analyze and process the electron microscope image to identify the components that match preset morphological features.

[0136] If the test sample contains neither characteristic elements nor components matching the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area to obtain a Raman spectrum; the processor analyzes the Raman spectrum to obtain the composition of the test sample.

[0137] In the process of analyzing the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range.

[0138] The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

[0139] In another embodiment of the present invention, a formulation raw material and excipient identification system based on multi-device combination is provided, including an electron microscope, an energy dispersive spectroscopy (EDS) spectrometer, an X-ray diffraction analysis system, an atomic force microscope (AFM) and a processor. In this embodiment, an X-ray diffraction analysis system is used instead of a Raman spectrometer, which can correlate electron microscopy, EDS, diffraction patterns, atomic force and other data to analyze the sample to be analyzed, and identify the components of the sample to be analyzed more accurately and quickly.

[0140] It should be noted that the above embodiments of the formulation raw material and excipient identification system based on multi-device combination and the formulation raw material and excipient identification method based on multi-device combination are based on the same inventive concept. All contents of the formulation raw material and excipient identification method based on multi-device combination are incorporated into the formulation raw material and excipient identification system based on multi-device combination by reference.

[0141] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0142] The above description is only a specific embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

[0143] The disclosure will now be described in conjunction with the various examples and combinations thereof described below.

[0144] 1. An example provides a method for identifying raw materials and excipients in a formulation based on multi-device collaboration, comprising the following steps:

[0145] A cross section of the test sample to be analyzed is subjected to cryo-polishing to obtain the test section;

[0146] The cross section to be tested is observed using an electron microscope to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested;

[0147] The test area was scanned using an energy dispersive spectrometer to obtain an energy spectrum image;

[0148] The electron microscope image and the energy dispersive spectroscopy image are used to determine whether the test sample contains characteristic elements. If characteristic elements are present, the characteristic elements are identified based on the energy dispersive spectroscopy image.

[0149] Based on the electron microscope image and the energy dispersive spectroscopy image, it is determined whether the test sample contains components that match the preset morphological features. If there are components that match the preset morphological features, the electron microscope image is analyzed and processed using a pre-built convolutional neural network model to identify the components that match the preset morphological features.

[0150] 2. The method for identifying raw materials and excipients in a formulation based on multi-device operation as described in Example 1 further includes the following steps:

[0151] If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area.

[0152] The composition of the test sample was obtained based on the Raman spectral analysis.

[0153] 3. According to the formulation raw material and excipient identification method based on multi-device combination provided in any of the foregoing examples, in the process of analyzing the composition of the test sample according to the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range.

[0154] The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

[0155] 4. The formulation raw material and excipient identification method based on multi-device combination provided in any of the foregoing examples further includes using a pre-built deep learning model to analyze and process the energy spectrum image to identify the feature elements, wherein the deep learning model is configured to perform feature extraction, noise reduction and image enhancement processing on the energy spectrum image to identify and display the feature elements.

[0156] 5. According to the formulation raw material and excipient identification method based on multi-device combination provided in any of the foregoing examples, the electron microscopy image is analyzed and processed using a pre-built convolutional neural network model to identify components matching preset morphological features, including:

[0157] The electron microscope image is preprocessed to obtain a processed electron microscope image. The preprocessing includes one or more of denoising, normalization, and size adjustment.

[0158] The processed electron microscope image is input into the convolutional neural network model. The convolutional neural network model predicts and segments the components of the processed electron microscope image that match the preset morphological features based on the morphological features of various polymer compounds that have been pre-annotated and learned.

[0159] 6. Based on the formulation raw material and excipient identification method based on multi-device use provided in any of the foregoing examples, the convolutional neural network model is pre-constructed in the following manner:

[0160] A learning sample set is obtained, which includes multiple learning samples. Each learning sample includes one or more electron microscope image samples corresponding to a known component. The known component has regular morphological features, and each electron microscope image sample is labeled with its corresponding known component.

[0161] A basic neural network model is determined, the learning sample set is input into the basic neural network model, and the basic neural network model is trained using a preset loss function to obtain the convolutional neural network model.

[0162] 7. Based on the multi-device-based formulation raw material and excipient identification method provided in any of the foregoing examples, predicting and segmenting components in the processed electron microscopy image that match preset morphological features to determine the components of each matched preset morphological feature, including:

[0163] Determine the substances corresponding to the components of each pre-defined morphological feature;

[0164] Obtain the segmented images corresponding to the components of each preset morphological feature. Each segmented image is configured to display a component of a preset morphological feature.

[0165] 8. The method for identifying raw materials and excipients in a formulation based on multi-device operation provided in any of the foregoing examples further includes:

[0166] The electron microscope image of the test section obtained by observing the test section with an electron microscope is configured as the first electron microscope image;

[0167] Based on the electron microscope image and the energy dispersive spectroscopy image, determine whether there are morphological features in the test sample. If morphological features are present, analyze and process the electron microscope image using a pre-built convolutional neural network model to identify components that match preset morphological features, including the following steps:

[0168] The test area is observed using an electron microscope to obtain a second electron microscope image of the test area, wherein the resolution of the second electron microscope image is greater than that of the first electron microscope image;

[0169] The second electron microscope image is then analyzed and processed using a pre-built convolutional neural network model to identify components that match preset morphological features.

[0170] 9. The method for identifying raw materials and excipients in a formulation based on multi-device combination provided in any of the foregoing examples, after analyzing and processing the electron microscopy image using a pre-built convolutional neural network model to identify components matching preset morphological features, further includes:

[0171] The energy spectrum image is further analyzed and processed using a pre-built deep learning model to identify components with similar morphological features.

[0172] 10. Based on the multi-device-based formulation raw material and excipient identification method provided in any of the foregoing examples, determine whether the test sample contains a component with a preset morphological characteristic by means of the following:

[0173] A morphological feature component library is constructed by pre-determining multiple components with regular morphological features. The morphological feature component library includes multiple components and reference morphological features corresponding to each component. The reference morphological features can be one or more.

[0174] The electron microscope image and the energy dispersive spectroscopy image are analyzed and compared with the morphological feature component library. If the morphological features of the component to be analyzed in the electron microscope image and / or the energy dispersive spectroscopy image match the reference morphological features, then it is determined that the test sample has a component with a matching preset morphological features.

[0175] 11. The formulation raw material and excipient identification method based on multi-device combination provided in any of the foregoing examples is applicable to analytical objects that include at least: formulations, polymer materials, and semiconductor materials; and / or,

[0176] It also includes the following steps:

[0177] If the test sample contains neither characteristic elements nor components matching the preset morphological characteristics, then X-ray diffraction analysis is used to analyze the composition of the test sample.

[0178] 12. Based on the multi-device-based method for identifying raw materials and excipients in a formulation provided in any of the foregoing examples, a multi-device-based system for identifying raw materials and excipients in a formulation is provided, including an electron microscope, an energy dispersive spectrometer, and a processor;

[0179] The formulation raw material and excipient identification system is configured to identify raw materials and excipients in the formulation in the following ways:

[0180] A cross section of the test sample to be analyzed is cryopolished to obtain the test section;

[0181] The electron microscope is used to observe the cross section to be tested to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested;

[0182] The energy dispersive spectrometer is used to scan the area to be tested to obtain an energy spectrum image;

[0183] The processor determines whether the test sample contains characteristic elements based on the electron microscope image and the energy dispersive spectroscopy image. If characteristic elements are present, the processor identifies the characteristic elements based on the energy dispersive spectroscopy image.

[0184] The processor determines whether the test sample contains components that match preset morphological features based on the electron microscope image and the energy dispersive spectroscopy image. If there are components that match preset morphological features, the processor uses a pre-built convolutional neural network model to analyze and process the electron microscope image to identify the components that match preset morphological features.

[0185] 13. The formulation raw material and excipient identification system based on multi-device combination provided in Example 12 also includes a Raman spectrometer;

[0186] The formulation raw material and excipient identification system is also configured to identify raw materials and excipients in the formulation in the following ways:

[0187] If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area.

[0188] The processor obtains the composition of the test sample based on the Raman spectrum analysis.

[0189] 14. The formulation raw material and excipient identification system based on multi-device combination provided in any of the foregoing examples further includes an atomic force microscope;

[0190] The formulation raw material and excipient identification system is also configured to identify raw materials and excipients in the formulation in the following ways:

[0191] In the process of analyzing the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range.

[0192] The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

Claims

1. A method for identifying raw materials and excipients in a pharmaceutical preparation based on multi-device collaboration, characterized in that, Includes the following steps: A cross section of the test sample to be analyzed is subjected to cryo-polishing to obtain the test section; The cross section to be tested is observed using an electron microscope to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested; The test area was scanned using an energy dispersive spectrometer to obtain an energy spectrum image; The electron microscope image and the energy dispersive spectroscopy image are used to determine whether the test sample contains characteristic elements. If characteristic elements are present, the characteristic elements are identified based on the energy dispersive spectroscopy image. Based on the electron microscope image and the energy dispersive spectroscopy image, it is determined whether the test sample contains components that match the preset morphological features. If there are components that match the preset morphological features, the electron microscope image is analyzed and processed using a pre-built convolutional neural network model to identify the components that match the preset morphological features.

2. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, It also includes the following steps: If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area. The composition of the test sample was obtained based on the Raman spectral analysis.

3. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 2, characterized in that, In the process of analyzing the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range. The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

4. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, It also includes using a pre-built deep learning model to analyze and process the energy spectrum image to identify the feature elements, wherein the deep learning model is configured to perform feature extraction, noise reduction, and image enhancement processing on the energy spectrum image to identify and display the feature elements.

5. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, The electron microscope images are analyzed and processed using a pre-built convolutional neural network model to identify components that match preset morphological features, including: The electron microscope image is preprocessed to obtain a processed electron microscope image. The preprocessing includes one or more of denoising, normalization, and size adjustment. The processed electron microscope image is input into the convolutional neural network model. The convolutional neural network model predicts and segments the components of the processed electron microscope image that match the preset morphological features based on the morphological features of various polymer compounds that have been pre-annotated and learned.

6. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 5, characterized in that, The convolutional neural network model is pre-built using the following method: A learning sample set is obtained, which includes multiple learning samples. Each learning sample includes one or more electron microscope image samples corresponding to a known component. The known component has regular morphological features, and each electron microscope image sample is labeled with its corresponding known component. A basic neural network model is determined, the learning sample set is input into the basic neural network model, and the basic neural network model is trained using a preset loss function to obtain the convolutional neural network model.

7. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 5, characterized in that, The components of the pre-defined morphological features in the processed electron microscope image are predicted and segmented to determine the components of each pre-defined morphological feature, including: Determine the substances corresponding to the components of each pre-defined morphological feature; Obtain the segmented images corresponding to the components of each preset morphological feature. Each segmented image is configured to display a component of a preset morphological feature.

8. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, Also includes: The electron microscope image of the test section obtained by observing the test section with an electron microscope is configured as the first electron microscope image; Based on the electron microscope image and the energy dispersive spectroscopy image, determine whether there are morphological features in the test sample. If morphological features are present, analyze and process the electron microscope image using a pre-built convolutional neural network model to identify components that match preset morphological features, including the following steps: The test area is observed using an electron microscope to obtain a second electron microscope image of the test area, wherein the resolution of the second electron microscope image is greater than that of the first electron microscope image; The second electron microscope image is then analyzed and processed using a pre-built convolutional neural network model to identify components that match preset morphological features.

9. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, For the test sample containing different components with matching morphological features, after analyzing the electron microscopy image using a pre-built convolutional neural network model to identify components matching preset morphological features, the method further includes: The energy spectrum image is further analyzed and processed using a pre-built deep learning model to identify components with similar morphological features.

10. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, The following method is used to determine whether the test sample contains components that match the preset morphological characteristics: A morphological feature component library is constructed by pre-determining multiple components with regular morphological features. The morphological feature component library includes multiple components and reference morphological features corresponding to each component. The reference morphological features can be one or more. The electron microscope image and the energy dispersive spectroscopy image are analyzed and compared with the morphological feature component library. If the morphological features of the component to be analyzed in the electron microscope image and / or the energy dispersive spectroscopy image match the reference morphological features, then it is determined that the test sample has a component with a matching preset morphological features.

11. The method for identifying raw materials and excipients in a formulation based on multi-device operation according to claim 1, characterized in that, The analytical objects applicable to multi-device-based methods for identifying raw materials and excipients within pharmaceutical preparations include at least: pharmaceutical preparations, polymeric materials, and semiconductor materials; and / or, It also includes the following steps: If the test sample contains neither characteristic elements nor components matching the preset morphological characteristics, then X-ray diffraction analysis is used to analyze the composition of the test sample.

12. A formulation raw material and excipient identification system based on multi-device operation, characterized in that, Includes electron microscopes, energy dispersive spectrometers, and processors; The formulation raw material and excipient identification system is configured to identify raw materials and excipients in the formulation in the following ways: A cross section of the test sample to be analyzed is cryopolished to obtain the test section; The electron microscope is used to observe the cross section to be tested to obtain an electron microscope image of the cross section to be tested, and a test area is determined on the cross section to be tested; The energy dispersive spectrometer is used to scan the area to be tested to obtain an energy spectrum image; The processor determines whether the test sample contains characteristic elements based on the electron microscope image and the energy dispersive spectroscopy image. If characteristic elements are present, the processor identifies the characteristic elements based on the energy dispersive spectroscopy image. The processor determines whether the test sample contains components that match preset morphological features based on the electron microscope image and the energy dispersive spectroscopy image. If there are components that match preset morphological features, the processor uses a pre-built convolutional neural network model to analyze and process the electron microscope image to identify the components that match preset morphological features.

13. The formulation raw material and excipient identification system based on multi-device operation according to claim 12, characterized in that, It also includes Raman spectrometers; The formulation raw material and excipient identification system is also configured to identify raw materials and excipients in the formulation in the following ways: If the test sample contains neither characteristic elements nor components that match the preset morphological characteristics, Raman spectroscopy is used to collect Raman spectra of the test area. The processor obtains the composition of the test sample based on the Raman spectrum analysis.

14. The formulation raw material and excipient identification system based on multi-device operation according to claim 13, characterized in that, It also includes atomic force microscopy; The formulation raw material and excipient identification system is also configured to identify raw materials and excipients in the formulation in the following ways: In the process of analyzing the composition of the test sample based on the Raman spectrum, if there are components with similar characteristic peaks, they are identified as target components. The presence of components with similar characteristic peaks is determined by the following method: the difference between the peak values ​​of at least two characteristic peaks is within a preset peak difference threshold range. The microstructure information of the test sample was obtained by atomic force microscopy analysis, and the target component was determined by combining the mechanical properties of the target component.

Images