1 Introduction

Teledental systems are developed to serve the patients from distant locations in this post-COVID context. Patients are advised not to visit the dental clinics to avoid the corona virus attacks. More they stay at home, more they are safe from COVID attacks. Security parameter is the foremost criterion that needs to be addressed while transmitting information in Teledental systems. In classical cryptography [1,2,3], the information is accessed by the -intruders in context of sniphing, spoofing, phising, etc. Symmetric key cryptography [4] in Teledental system means the same key will be used by the patients and physicians for encryption and decryption respectively. Confidential data related to the patients need to be transmitted to the destination by using certain encryption technique. It is so done in order to resist the man in the middle attacks. Such encryption and decryption procedures are controlled by the concerned Teledental system. In medical sciences, data transmission using artificial neural architectures is an emerging area of research [5].

Tampering of dental data by the intruders lead to wrong diagnosis and failure in line of treatments. Unprotected data can easily be altered intentionally to leak out the information. If the symmetric key used for encryption is somehow compromised by the attackers then the entire patients’ data will be leaked to the unauthorized agents. The implementation of artificial neural networks (ANN) [6, 7] is a secure option for data transmission inside the Telecare E-Health system [8, 9]. A perceptron is a type of artificial neural network which has been considered in this proposed technique. It consists of input neurons, weights and a bias constant connected hidden neurons, and output neuron. The key idea is that two perceptrons are placed at the patient and dentist, and they do not exchange their keys between them. Learning rules are used to converge their weight vector into an identical session key. The intraoral information is broken into four parts; perceptron at the patient end is followed by a demultiplexer of size 1 × 4 (Fig. 1).

Fig. 1
figure 1

Single layer feed forward perceptron

Full size image

Proper oral care is always needed for everyone’s healthy lifestyle. Gingivitis is a kind of periodontal disease. Its symptoms are gum swelling, bleeding, redness, soreness, etc. Deposition of bacterial plaque on the teeth surface is the initial symptom of such disease. Then the plaque is calcified to form tartar and thus causing the gingivitis. Gingivitis can be reversed in case of early detection provided the patient takes a constant and better oral hygiene. The best way to achieve it by a proper brushing technique and minimum twice daily [10,11,12,13]. In the advanced Telecare E-Health, the patients transmit her/his intraoral data to the specified dentist for diagnosis and treatment.

This paper presents an encryption technique based on perceptron that transmit the intraoral data by dismantling into sub-pieces of square matrix by devising through demultiplexer by the patient. Then encrypting with synchronized session key followed by header generation. The reverse end perceptron will decrypt the header portions from each share followed by decryption with synchronized session key one by one, then feeding into a multiplexer to reconstruct the original intraoral data. It deals with low computational cost and complexity incurred by the perceptron based neural network.

A secured encryption technique has been proposed to share the intraoral data by applying demultiplexing technique in the domain of Telecare E-Health. Aiming to make the intruders’ task more difficult, a header generation has been proposed at each share. This paper has been organized in the following manner. Section 1 deals with the introductory part. Section 2 focuses some of the background and related tasks. Section 3 contains the problematic areas while transmitting data in Telecare systems. The novelty of the proposed technique has been illustrated in Sect. 4. Results and analysis are given in Sect. 5. Section 6 deals with the conclusion and those references are mentioned at the end.

2 Literature survey

2.1 Causes of gingivitis

Food habits and lifestyles are also responsible for gingivitis. The most common causes of gingivitis are poor oral hygiene which encourages the formation of plaque. Plaque is also formed when the starches and sugar in food interact with the oral bacterial microflora. Brushing and flossing each day removes plaque. Risk factors include certain viral and fungal infections, poor nutrition, older age, hormonal changes, etc. [14].

Gum diseases are being indicated by the early signs of gum bleeding. If treated early and on the right track, there is a possibility of reversing the conditions [15]. Gingivitis is certainly connected to diabetes and other related chronic diseases. A Fig. 2 depicts the image of gingivitis caused due to diabetes. In India, huge number of people suffers from diabetes, which leads to gingivitis in their lifespan [16, 17].

Fig. 2
figure 2

Gingivitis caused due to diabetes

Full size image

Other factors contributing largely to gingivitis are hormonal changes, stress and anxiety, malnutrition, some medications like phenytoin, amlodipine, cyclosporine, diabetes, immune compromization, dental caries, etc.

2.2 Use of perceptron in data transmission

In the field of Telcare E-Health, huge amount of extensive data are available for the research and transaction purposes [18,19,20]. So it is necessary to propose a model for the efficient data filtration and transmission. In the same regard, arrival of artificial neural networks has brought a higher degree of pace in medical data transmission [21,22,23]. An artificial neural network may be represented as a complex system [24] with higher dimensions, extensive interconnections, adaptability, and self-organizing features. Single layered and double layered feed forward neural networks were very useful in the recent decades in the domain of machine learning and big data analysis [23, 25]. A special type of feed forward neural network is the multilayer perceptron (MLP) [26, 27] having multiple hidden layers is more efficient to solve different types of nonlinear problems like pattern recognition, big data classification, numerical approximation, etc. Earlier research study had shown that single layer perceptrons outperform any continuous or discontinuous function with compared with multi layer perceptron [28]. Hence single layer perceptron are much better in data transmission domain. Mirjalili had shown that in the context of the back propagation neural networks, the starting values of the learning rate and the momentum factor lead to unsuitable divergence [29].

3 Problem specifications

The medical travelling throughout the Telecare network are the most vulnerable parts by the intruders. They are very much proactive to sense those patients’ related confidential data, and then to forged them entirely. In Telecare E-Health systems, patient can take active participation from their home regarding their health issues. Similarly, dental issues can be addressed by such Telecare systems [5]. Despite of rapid developments in this sector, there are loopholes in those systems in the light of data security. An encryption algorithm is being considered as good as its keys are robust and sensitive. Moreover, the used encryption and decryption algorithms are open accessible by all agents. Only the key is kept secret between the patient and the dentist. Hence, by compromising the secret key, intruders are able to detect all confidential data. This is the area of concern in this proposed technique.

4 Proposed methodology

The key issue as stated in the earlier section has been addressed in this proposed paper. Structurally similar perceptrons are taken at the patient’s and the dentist’s Telecare system [8, 9]. The input set and the random weight vectors are same at both ends to achieve the synchronization. Based on the leaning rules of the perceptron, will adjust its weights till the best condition is not reached. Thus, in true sense, there will be no exchange of the secret key between the patient and dentist. Followed by the demultiplexing technique on the intraoral data, then a header generation has been proposed on all the partial shares [5]. The following algorithm will explain its technique in brief.

figure a

4.1 Proposed header generation

Another novel part which has been proposed in this paper is to design a header structure for individual shares [5]. It consists of two attributes: quadrant number (QNo) and the total length of the partial share, which is dynamic in size. It depends on the length of the partial share generated by demultiplexer. The quadrant number is used for the reassembly by the dentist’s perceptron. The following algorithm will illustrate the header generation.

figure b

5 Results analysis

An extensive analysis has been done on the proposed technique of encryption. The comparison study was carried out between the proposed technique and the existing RSA, 3-DES (\(168\) bits), AES (\(128\) bits) on thirty different types of files related to intraoral data. After compiling such files, a comparison which is emphasized on Avalanche test, Strict Avalanche tests and Bit Independence test has been carried out. Avalanche Test, Strict Avalanche Test and Bit Independence Test.

The study of comparison between the source intraoral file and proposed encrypted intraoral file to observe the magnitude of changes in its all bits has been done. Change in the encrypted bytes has been observed for every bit change with respect to the original intraoral byte in the entire sequence of communication in Telecare E-Health. Here, the standard deviation from the expected values were calculated and then subtract the ratio of the calculated standard deviation with expected value from 1.0 to get the Avalanche and Strict Avalanche on a 0.0–1.0 scale. The Bit Independence test measures the correlation coefficient between the \(j\)th and \(k\)th components of the output difference string. If this correlation gets closer to \(1.0\) then it is said to have good Avalanche and Strict Avalanche criteria. In case of intraoral text files, there will be zero bytes between \(128\) and \(255\) bytes. The function of Bit Independence test can be written in the following Eq. 1:

$$f : \left\{ {0,1} \right\}^{n} \left\{ {0,1} \right\}^{n} ,$$
(1)

for all i, j, k \(\in {\{\mathrm{0,1}\}}^{n}\),\(j\ne k\), flipping input bit \(i\) will generate output bits \(j\) and \(k\) will change independently.

Thirty intraoral files of different sizes ranging from \(3296\) to \(\mathrm{5,456,704}\) bytes were considered for carrying out Avalanche test, Strict Avalanche test and Bit Independence test. The output of Avalanche test and Strict Avalanche test on those intraoral files encrypted using the proposed technique and existing techniques like RSA, 3DES, AES is almost equal to 1.0. The output of Bit Independence test is comparatively less than Avalanche test and Strict Avalanche test. Table 1 contains the observed data in this regard. In Fig. 3, a graphical representation has been put on the outputs listed in the above Table 1 for Avalanche test, Strict Avalanche test and Bit Independence test on their average data in the context of the proposed technique and existing techniques like RSA, 3DES and AES.

Table 1 Comparisons of average values on different files
Full size table
Fig. 3
figure 3

Character frequency of plain intraoral file

Full size image

5.1 Non-homogeneity test

To find out the significant difference between the expected frequency and the actual frequency of the encrypted intraoral data, Pearson’s Chi-square test has been used. Higher Chi-square value indicated better degree of encryption at the proposed technique. Thirty intraoral files of different sizes ranging from \(3296\) to 5,456,704 bytes were considered for performing Chi-square test. Table 2 represents Chi-square values obtained by the proposed technique, 3DES, and AES of thirty different intraoral files. The average Chi-square values obtained by the proposed technique, 3DES, and AES are 9,236,291, 9,227,496, and 10,453,517 respectively. There has been an increase in the Chi-square value when the intraoral file sizes increases. So, it may obtain better degree of security in proposed which is comparable with that of others.

Table 2 Comparisons of Chi-square average values
Full size table

5.2 Comparative tabulation with classical algorithms

In this section, a comparative study was conducted between the proposed technique and existing bench-mark classical 3DES and AES technique [30]. It has been observed that the proposed technique had yielded satisfactory results when compared with existing methods. The summary of this statement is given at Table 3.

Table 3 Comparative study w.r.t. existing algorithms
Full size table

5.3 Character frequencies, floating frequencies, autocorrelation analysis

In this section the analysis of character frequencies of thirty intraoral files were carried out examination by the proposed technique. The following Fig. 3 represents the spectrum which contains the frequency distribution of its characters for a single intraoral file. In the next Fig. 4, it contains the spectrum of its frequency distribution of the corresponding encrypted characters by the proposed method when applied on the same intraoral file. From these two figures, it can be stated that that the frequency distribution of characters are more widely distributed in the encrypted intraoral file [31]. The floating frequency analysis of thirty intraoral files was carried out here by the proposed technique. It is a characteristic of the local information content for the intraoral files at different points. The floating frequency indicates how many different characters of sixty four character long segment inside the source file. Figure 5 depicts the floating frequencies of its characters for a single intraoral file. The Fig. 6 indicates the spectrum of floating frequencies of the corresponding encrypted characters by using the proposed technique on the same intraoral file. By observing these two figures, it can be stated that floating frequencies of by the proposed technique shows higher degree of security. Analysis of autocorrelation of thirty different types of intraoral files was done by the proposed technique. Autocorrelation of a file is an index of the similarity at different points inside it. The spectrum of autocorrelation of characters when applied on the same intraoral file has been shown in the following Fig. 7. The spectrum of autocorrelation of encrypted file using the proposed technique can be found in Fig. 8. Enhanced autocorrelation has been observed in this proposed technique.

Fig. 4
figure 4

Character frequency of plain intraoral file by proposed encryption

Full size image
Fig. 5
figure 5

Floating frequency of plain intraoral file

Full size image
Fig. 6
figure 6

Floating frequency of plain intraoral file by proposed encryption

Full size image
Fig. 7
figure 7

Autocorrelation of plain intraoral file

Full size image
Fig. 8
figure 8

Autocorrelation of plain intraoral file by proposed encryption

Full size image

5.4 Performance comparison with classical encryption

As we know that IDEA algorithm works on 64 bits plaintext to 64 bits cipher text by using 128 bits secret key, but the proposed technique offers the same on n bits plaintext to n bits cipher text with key generated by the perceptrons at the patient and dentist end. The most powerful linear cryptanalysis attack was done on the DES algorithm due to availability of known plaintext. The proposed technology is fully dependent on the intraoral files of different patients, which are by far different from one another. RC5 is basically word oriented encryption where the length of the word may be 16 bits, 32 bits, 64 bits. The proposed technique allows generating a header addition to each partial share. Through this strategy, the better performance of the proposed technique has been represented [32].

6 Conclusions

In this Telecare E-Health field, patients’ medical data can be compromised by different types of attacks on the system. To ensure that this paper has proposed an efficient and secured technique to transmit the intraoral data. Symmetric key has been used her which is generated by the perceptron neural machine. The advantage is that there needs no secret key exchange between the patients and the dentist for symmetric cryptography. Statistically, the proposed technique has been shown as an efficient technique for data transmission in the sector of Telecare dental E-Health.